Educating national security leaders on artificial intelligence

Understanding artificial intelligence and how it relates to matters of national security has become a top priority for military and government leaders in recent years. A new three-day custom program entitled “Artificial Intelligence for National Security Leaders” — AI4NSL for short — aims to educate leaders who may not have a technical background on the basics of AI, machine learning, and data science, and how these topics intersect with national security.

“National security fundamentally is about two things: getting information out of sensors and processing that information. These are two things that AI excels at. The AI4NSL class engages national security leaders in understanding how to navigate the benefits and opportunities that AI affords, while also understanding its potential negative consequences,” says Aleksander Madry, the Cadence Design Systems Professor at MIT and one of the course’s faculty directors.

Organized jointly by MIT’s School of Engineering, MIT Stephen A. Schwarzman College of Computing, and MIT Sloan Executive Education, AI4NSL wrapped up its fifth cohort in April. The course brings leaders from every branch of the U.S. military, as well as some foreign military leaders from NATO, to MIT’s campus, where they learn from faculty experts on a variety of technical topics in AI, as well as how to navigate organizational challenges that arise in this context.

“We set out to put together a real executive education class on AI for senior national security leaders,” says Madry. “For three days, we are teaching these leaders not only an understanding of what this technology is about, but also how to best adopt these technologies organizationally.”

The original idea sprang from discussions with senior U.S. Air Force (USAF) leaders and members of the Department of the Air Force (DAF)-MIT AI Accelerator in 2019.

According to Major John Radovan, deputy director of the DAF-MIT AI Accelerator, in recent years it has become clear that national security leaders needed a deeper understanding of AI technologies and its implications on security, warfare, and military operations. In February 2020, Radovan and his team at the DAF-MIT AI Accelerator started building a custom course to help guide senior leaders in their discussions about AI.

“This is the only course out there that is focused on AI specifically for national security,” says Radovan. “We didn’t want to make this course just for members of the Air Force — it had to be for all branches of the military. If we are going to operate as a joint force, we need to have the same vocabulary and the same mental models about how to use this technology.”

After a pilot program in collaboration with MIT Open Learning and the MIT Computer Science and Artificial Intelligence Laboratory, Radovan connected with faculty at the School of Engineering and MIT Schwarzman College of Computing, including Madry, to refine the course’s curriculum. They enlisted the help of colleagues and faculty at MIT Sloan Executive Education to refine the class’s curriculum and cater the content to its audience. The result of this cross-school collaboration was a new iteration of AI4NSL, which was launched last summer.

In addition to providing participants with a basic overview of AI technologies, the course places a heavy emphasis on organizational planning and implementation.

“What we wanted to do was to create smart consumers at the command level. The idea was to present this content at a higher level so that people could understand the key frameworks, which will guide their thinking around the use and adoption of this material,” says Roberto Fernandez, the William F. Pounds Professor of Management and one of the AI4NSL instructors, as well as the other course’s faculty director.

During the three-day course, instructors from MIT’s Department of Electrical Engineering and Computer Science, Department of Aeronautics and Astronautics, and MIT Sloan School of Management cover a wide range of topics.

The first half of the course starts with a basic overview of concepts including AI, machine learning, deep learning, and the role of data. Instructors also present the problems and pitfalls of using AI technologies, including the potential for adversarial manipulation of machine learning systems, privacy challenges, and ethical considerations.

In the middle of day two, the course shifts to examine the organizational perspective, encouraging participants to consider how to effectively implement these technologies in their own units.

“What’s exciting about this course is the way it is formatted first in terms of understanding AI, machine learning, what data is, and how data feeds AI, and then giving participants a framework to go back to their units and build a strategy to make this work,” says Colonel Michelle Goyette, director of the Army Strategic Education Program at the Army War College and an AI4NSL participant.

Throughout the course, breakout sessions provide participants with an opportunity to collaborate and problem-solve on an exercise together. These breakout sessions build upon one another as the participants are exposed to new concepts related to AI.

“The breakout sessions have been distinctive because they force you to establish relationships with people you don’t know, so the networking aspect is key. Any time you can do more than receive information and actually get into the application of what you were taught, that really enhances the learning environment,” says Lieutenant General Brian Robinson, the commander of Air Education and Training Command for the USAF and an AI4NSL participant.

This spirit of teamwork, collaboration, and bringing together individuals from different backgrounds permeates the three-day program. The AI4NSL classroom not only brings together national security leaders from all branches of the military, it also brings together faculty from three schools across MIT.

“One of the things that's most exciting about this program is the kind of overarching theme of collaboration,” says Rob Dietel, director of executive programs at Sloan School of Management. “We're not drawing just from the MIT Sloan faculty, we're bringing in top faculty from the Schwarzman College of Computing and the School of Engineering. It's wonderful to be able to tap into those resources that are here on MIT’s campus to really make it the most impactful program that we can.”

As new developments in generative AI, such as ChatGPT, and machine learning alter the national security landscape, the organizers at AI4NSL will continue to update the curriculum to ensure it is preparing leaders to understand the implications for their respective units.

“The rate of change for AI and national security is so fast right now that it's challenging to keep up, and that's part of the reason we've designed this program. We've brought in some of our world-class faculty from different parts of MIT to really address the changing dynamic of AI,” adds Dietel.

de MIT News https://ift.tt/uBXRcoU

Researchers teach an AI to write better chart captions

Chart captions that explain complex trends and patterns are important for improving a reader’s ability to comprehend and retain the data being presented. And for people with visual disabilities, the information in a caption often provides their only means of understanding the chart.

But writing effective, detailed captions is a labor-intensive process. While autocaptioning techniques can alleviate this burden, they often struggle to describe cognitive features that provide additional context.

To help people author high-quality chart captions, MIT researchers have developed a dataset to improve automatic captioning systems. Using this tool, researchers could teach a machine-learning model to vary the level of complexity and type of content included in a chart caption based on the needs of users.

The MIT researchers found that machine-learning models trained for autocaptioning with their dataset consistently generated captions that were precise, semantically rich, and described data trends and complex patterns. Quantitative and qualitative analyses revealed that their models captioned charts more effectively than other autocaptioning systems.  

The team’s goal is to provide the dataset, called VisText, as a tool researchers can use as they work on the thorny problem of chart autocaptioning. These automatic systems could help provide captions for uncaptioned online charts and improve accessibility for people with visual disabilities, says co-lead author Angie Boggust, a graduate student in electrical engineering and computer science at MIT and member of the Visualization Group in the Computer Science and Artificial Intelligence Laboratory (CSAIL).

“We’ve tried to embed a lot of human values into our dataset so that when we and other researchers are building automatic chart-captioning systems, we don’t end up with models that aren’t what people want or need,” she says.

Boggust is joined on the paper by co-lead author and fellow graduate student Benny J. Tang and senior author Arvind Satyanarayan, associate professor of computer science at MIT who leads the Visualization Group in CSAIL. The research will be presented at the Annual Meeting of the Association for Computational Linguistics.

Human-centered analysis

The researchers were inspired to develop VisText from prior work in the Visualization Group that explored what makes a good chart caption. In that study, researchers found that sighted users and blind or low-vision users had different preferences for the complexity of semantic content in a caption. 

The group wanted to bring that human-centered analysis into autocaptioning research. To do that, they developed VisText, a dataset of charts and associated captions that could be used to train machine-learning models to generate accurate, semantically rich, customizable captions.

Developing effective autocaptioning systems is no easy task. Existing machine-learning methods often try to caption charts the way they would an image, but people and models interpret natural images differently from how we read charts. Other techniques skip the visual content entirely and caption a chart using its underlying data table. However, such data tables are often not available after charts are published.

Given the shortfalls of using images and data tables, VisText also represents charts as scene graphs. Scene graphs, which can be extracted from a chart image, contain all the chart data but also include additional image context.

“A scene graph is like the best of both worlds — it contains almost all the information present in an image while being easier to extract from images than data tables. As it’s also text, we can leverage advances in modern large language models for captioning,” Tang explains.

They compiled a dataset that contains more than 12,000 charts — each represented as a data table, image, and scene graph — as well as associated captions. Each chart has two separate captions: a low-level caption that describes the chart’s construction (like its axis ranges) and a higher-level caption that describes statistics, relationships in the data, and complex trends.

The researchers generated low-level captions using an automated system and crowdsourced higher-level captions from human workers.

“Our captions were informed by two key pieces of prior research: existing guidelines on accessible descriptions of visual media and a conceptual model from our group for categorizing semantic content. This ensured that our captions featured important low-level chart elements like axes, scales, and units for readers with visual disabilities, while retaining human variability in how captions can be written,” says Tang.

Translating charts

Once they had gathered chart images and captions, the researchers used VisText to train five machine-learning models for autocaptioning. They wanted to see how each representation — image, data table, and scene graph — and combinations of the representations affected the quality of the caption.

“You can think about a chart captioning model like a model for language translation. But instead of saying, translate this German text to English, we are saying translate this ‘chart language’ to English,” Boggust says.

Their results showed that models trained with scene graphs performed as well or better than those trained using data tables. Since scene graphs are easier to extract from existing charts, the researchers argue that they might be a more useful representation.

They also trained models with low-level and high-level captions separately. This technique, known as semantic prefix tuning, enabled them to teach the model to vary the complexity of the caption’s content.

In addition, they conducted a qualitative examination of captions produced by their best-performing method and categorized six types of common errors. For instance, a directional error occurs if a model says a trend is decreasing when it is actually increasing.

This fine-grained, robust qualitative evaluation was important for understanding how the model was making its errors. For example, using quantitative methods, a directional error might incur the same penalty as a repetition error, where the model repeats the same word or phrase. But a directional error could be more misleading to a user than a repetition error. The qualitative analysis helped them understand these types of subtleties, Boggust says.

These sorts of errors also expose limitations of current models and raise ethical considerations that researchers must consider as they work to develop autocaptioning systems, she adds.

Generative machine-learning models, such as those that power ChatGPT, have been shown to hallucinate or give incorrect information that can be misleading. While there is a clear benefit to using these models for autocaptioning existing charts, it could lead to the spread of misinformation if charts are captioned incorrectly.

“Maybe this means that we don’t just caption everything in sight with AI. Instead, perhaps we provide these autocaptioning systems as authorship tools for people to edit. It is important to think about these ethical implications throughout the research process, not just at the end when we have a model to deploy,” she says.

Boggust, Tang, and their colleagues want to continue optimizing the models to reduce some common errors. They also want to expand the VisText dataset to include more charts, and more complex charts, such as those with stacked bars or multiple lines. And they would also like to gain insights into what these autocaptioning models are actually learning about chart data.

This research was supported, in part, by a Google Research Scholar Award, the National Science Foundation, the MLA@CSAIL Initiative, and the United States Air Force Research Laboratory.

de MIT News https://ift.tt/eO3A25L

Transatlantic connections make the difference for MIT Portugal

Successful relationships take time to develop, with both parties investing energy and resources and fostering mutual trust and understanding. The MIT Portugal Program (MPP), a strategic partnership between MIT, Portuguese universities and research institutions, and the Portuguese government, is a case in point.

Portugal’s inaugural partnership with a U.S. university, MPP was established in 2006 as a collaboration between MIT and the Portuguese Science and Technology Foundation (Fundação para a Ciência e Tecnologia, or FCT). Since then, the program has developed research platforms in areas such as bioengineering, sustainable energy, transportation systems, engineering design, and advanced manufacturing. Now halfway through its third phase (MPP2030, begun in 2018), the program owes much of its success to the bonds connecting institutions and people across the Atlantic over the past 17 years.

“When you look at the successes and the impact, these things don’t happen overnight,” says John Hansman, the T. Wilson Professor of Aeronautics and Astronautics at MIT and co-director of MPP, noting, in particular, MPP’s achievements in the areas of energy and ocean research, as well as bioengineering. “This has been a longstanding relationship that we have and want to continue. I think it’s been beneficial to Portugal and to MIT. I think you can argue it has made substantial contributions to the success that Portugal is currently experiencing both in its technical capabilities and also its energy policy.”

With research often aimed at climate and sustainability solutions, one of MPP’s key strengths is its education of future leaders in science, technology, and entrepreneurship. And the program’s impacts carry forward, as several former MPP students are now on the faculty at participating Portuguese universities.

“The original intent of working together with Portugal was to try to establish collaboration between universities and to instill some of the MIT culture with the culture in Portugal, and I think that’s been hugely successful,” says Douglas Hart, MPP co-director and professor of mechanical engineering at MIT. “It has had a lot of impacts in terms of the research, but also the people.”

One of those people is André Pina, associate director of H2 strategy and origination at the company EDP, who was in residence at MIT in 2014 as part of the MPP Sustainable Energy Systems Doctoral Program. He says the competencies and experiences he acquired have been critical to his professional development in energy system planning, have influenced his approach to problem solving, and have allowed him to bring "holistic thinking" to business endeavors.

"The MIT Portugal Program has created a collaborative ecosystem between Portuguese universities, companies, and MIT that enabled the training of highly qualified professionals, while contributing to the positioning of Portuguese companies in new cutting-edge fields,” he says.

Building on MPP’s previous successes, MPP2030 focuses on advancing research in four strategic areas: climate science and climate change; earth systems from oceans to near space; digital transformation in manufacturing; and sustainable cities — all involving data science-intensive approaches and methodologies. Within these broad scientific areas, FCT funding has enabled seven collaborative large-scale “flagship” projects between Portuguese and MIT researchers during the current phase, as well as dozens of smaller projects.

Flagship projects currently underway include:

·   AEROS Constellation

·   C-Tech: Climate Driven Technologies for Low Carbon Cities

·   K2D: Knowledge and Data from the Deep to Space

·   NEWSAT

·   Operator: Digital Transformation in Industry with a Focus on the Operator 4.0

·   SNOB-5G: Scalable Network Backhauling for 5G

·   Transformer 4.0: Digital Revolution of Power Transformers

Sustainability plays a significant role in MPP — reflective of the value both Portugal and MIT place on environmental, energy, and climate solutions. Projects under the Sustainable Cities strategic area, for example, are “helping cities in Portugal to become more efficient and more sustainable,” Hansman says, noting that MPP’s influence is being felt in cities across the country and it is “having a big impact in terms of local city planning activities.”

Regarding energy, Hansman points to a previous MPP phase that focused on the Azores as an isolated energy ecosystem and investigated its ability to minimize energy use and become energy independent.

“That view of system-level energy use helped to stimulate activity on the mainland in Portugal, which has helped Portugal become a leader in various energy sources and made them less vulnerable in the last year or two,” Hansman says.

In the Oceans to Near Space strategic area, the K2D flagship project also emphasizes research into sustainability solutions, as well as resilience to environmental change. Over the past few years, K2D researchers in Portugal and MIT have worked together to develop components that permit cost-effective gathering of chemical, physical, biological, and environmental data from the ocean depths. One current project investigates the integration of autonomous underwater vehicles with subsea cables to enhance both environmental monitoring and hazard warning systems.

“The program has been very successful,” Hart says. “They are now deploying a 2-kilometer cable just south of Lisbon, which will be in place in another month or so. Portugal has been hit with tsunamis that caused tremendous devastation, and one of the objectives of these cables is to sense tsunamis. So, it’s an early warning system.”

As a leader in ocean technology with a long history of maritime discovery, Portugal provides many opportunities for MIT’s ocean researchers. Hart notes that the Portuguese military invites international researchers on board its ships, providing MIT with research opportunities that would be financially difficult otherwise.

Hansman adds that partnering with researchers in the Azores provides MIT with unique access to facilities and labs in the middle of the Atlantic Ocean. For example, Hart will be teaching at a marine robotics summer school in the Azores this July.

Cadence Payne, an MIT PhD candidate, is among those planning to attend. Through MPP’s AEROS project, Payne has helped develop a modular “cubesat” that will orbit over Portugal’s Exclusive Economic Zone collecting images and radio data to help define the ecological health of the country’s coastal waters. The nanosatellite is expected to launch in late 2023 or early 2024, says Payne, adding that it will be Portugal’s first cubesat mission.

“In monitoring the ocean, you’re monitoring the climate,” Payne says. “If you want to do work on detecting climate change and developing methods of mitigating climate change … it helps to integrate international collaboration,” she says, adding that, for students, “it’s been a really beautiful opportunity for us to see the benefits of collaboration.”

“I would say one of the main benefits of working with Portugal is that we share many interests in research in the sense that they’re very interested in climate change, sustainability, environmental impacts and those kinds of things,” says Hart. “They have turned out to be a very good strategic partner for MIT, and, hopefully, MIT for them.”

de MIT News https://ift.tt/Mt0n5jG

Summer 2023 recommended reading from MIT

Summer is the perfect time to curl up with a good book — and MIT authors have had much to offer in the past year. The following titles represent a selection of books published in the past 12 months by MIT faculty and staff. In addition to links for each book from its publisher, the MIT Libraries has compiled a helpful list of the titles held in its collections.

Looking for more literary works from the MIT community? Enjoy our book lists from 2022 and 2021.

Happy reading!

Novel and poetry

“The Study of Human Life” (Penguin, 2022)
By Joshua Bennett, visiting professor of literature

In this deeply personal book, Bennett, a visiting professor who joins the MIT faculty this summer, recalls and reimagines social worlds almost but not entirely lost, all while gesturing toward the ones we are building now, in the midst of a state of emergency, together. Bennett opens with a set of autobiographical poems with themes of family, life, death, vulnerability, and the joys and dreams of youth. The central section features an alternate history where Malcolm X is resurrected from the dead, as is a young Black man shot by police some 50 years later. The final section includes poems about fatherhood, on the heels of Bennett's first child being born.

“Pomegranate” (Atria Books, 2023)
By Helen Elaine Lee, professor in comparative media studies/writing

This novel from Helen Elaine Lee focuses on a queer Black woman working to stay clean, pull her life together, and heal after being released from prison. In lyrical and precise prose, Lee paints a humane and unflinching portrait of the devastating effects of incarceration and addiction, and of one woman’s determination to tell her story.

Science

“Responding to the Climate Threat: Essays on Humanity’s Greatest Challenge” (Springer Cham, 2023)
By Henry Jacoby, the William F. Pounds Professor Emeritus in management; Gary Yohe; Richard Richels; and Benjamin Santer

Four veteran climate experts present our current understanding of the climate threat and what can be done about it, in lay language ― without losing critical aspects of the natural and social science. In a series of essays, they explain the essential components of the challenge, countering the forces of distrust of the science and opposition to a vigorous national response.

“The Transcendent Brain: Spirituality in the Age of Science” (Pantheon, 2023)
By Alan Lightman, professor of the practice in humanities

Drawing on intellectual history and conversations with contemporary scientists, philosophers, and psychologists, Lightman asks a series of thought-provoking questions that illuminate our strange place between the world of particles and forces and the world of complex human experience.

“Climate Future: Averting and Adapting to Climate Change” (Oxford University Press, 2022)
By Robert S. Pindyck, the Bank of Tokyo-Mitsubishi Ltd Professor in Finance and Economics

Climate change initiatives typically focus on reducing greenhouse gas emissions — but what happens if these efforts fall short? In his latest book, Pindyck, an environmental economist, contends that most countries will not come close to meeting their CO2 reduction goals. He recommends adaptations such as sea walls and dykes, hybrid crops, and large-scale geoengineering.

Culture, humanities, and social sciences

“Spoken Word: A Cultural History” (Knopf, 2023)
By Joshua Bennett, visiting professor of literature

A celebration of voices outside the dominant cultural narrative, who boldly embraced an array of styles and forms and redefined what — and whom — the mainstream would include, Bennett’s book illuminates the profound influence spoken word has had everywhere melodious words are heard, from Broadway to academia, from the podiums of political protest to cafés, schools, and rooms full of strangers all across the world.

“A Thousand Steps to Parliament: Constructing Electable Women in Mongolia” (University of Chicago, 2022)
By Manduhai Buyandelger, professor of anthropology

This book traces how the complicated, contradictory paths to political representation that women in Mongolia must walk mirror those the world over. Buyandelger shows how successful women candidates use strategies of self-polishing to cultivate charisma and a reputation for being oyunlag, or intellectual. By tracing the complicated, contradictory paths to representation that women in Mongolia must walk, "A Thousand Steps to Parliament" holds a mirror up to democracies the world over, revealing an urgent need to grapple with the encroaching effects of neoliberalism in our global political systems.

“Lines Drawn across the Globe: Reading Richard Hakluyt’s ‘Principal Navigations’” (McGill-Queen's University Press, 2023)
By Mary C. Fuller, professor of literature

Around 1600, the English geographer and cleric Richard Hakluyt sought to honor his nation by publishing a compilation of every document he could find relating to its voyages and trade beyond the boundaries of Europe. Fuller traces the history of the book's compilation and gives order and meaning to its famously diverse contents: from Sierra Leone to Iceland, from Spanish narratives of New Mexico to French accounts of the Saint Lawrence and Portuguese accounts of China.

“Playing Oppression: The Legacy of Conquest and Empire in Colonialist Board Games” (MIT Press, 2023)
By Mikael Jakobsson, lecturer in comparative media studies, and Mary Flanagan

In "Playing Oppression," Jakobsson and Flanagan apply incisive frameworks of postcolonial theory to a broad historical survey of board games to show how seemingly benign entertainments reinforce the logic of imperialism. The work deftly analyzes this insidious violence and proposes a path forward with board games that challenge colonialist thinking and embrace a much broader cultural imagination.

“A New Companion to Herman Melville” (Wiley, 2022)
Edited by Wyn Kelley, senior lecturer of literature, and Christopher Ohge

Building on the success of the first Blackwell "Companion to Herman Melville," and offering a variety of tools for reading, writing, and teaching Melville, "A New Companion to Herman Melville" delivers an insightful examination of Melville for the 21st century. Editors Kelley and Ohge create a framework that reflects a pluralistic model for humanities teaching and research offering critical, technological, and aesthetic practices that can be employed to read Melville in exciting and revelatory ways.

“The Scarce State: Inequality and Political Power in the Hinterland” (Cambridge University Press, 2023)
By Noah L. Nathan, professor of political science

The formal state often appears absent in the rural periphery in developing countries. Yet these states are not as weak as many believe. In a multi-method study of historical development in Ghana, Nathan rethinks the process of state-building in hinterlands, demonstrating how even seemingly absent states still change the underlying nature of their societies, with implications for better understanding the governance challenges that these regions face. 

“Eating Beside Ourselves: Thresholds of Foods and Bodies” (Duke University Press, 2023)
Edited by Heather Paxson, the William R. Kenan, Jr. Professor of Anthropology and head of MIT Anthropology

“Eating beside Ourselves” examines eating as a site of transfer and transformation across bodies and selves. The contributors show that by turning organic substance into food, acts of eating create interconnected food webs organized by relative conditions of edibility through which eaters may in turn become eaten.

“Writing Egypt: Al-Maqrizi and His Historical Project” (Edinburgh University Press, 2022)
By Nasser Rabbat, professor of architecture

Although al-Maqrizi is recognized as the most influential historian of premodern Egypt, he has never received the probing historical treatment warranted by his standing and scholarly output. This book fills that gap. Arranged in three sections, it tells al-Maqrizi’s life story; weaves it with historiographical, textual, and methodological analysis of his oeuvre; and reconstructs the afterlife of the author and his work down to the present.

“Life Is Hard” (Riverhead Books, 2022)
By Kieran Setiya, professor of philosophy

There is no cure for the human condition: Life is hard. But Setiya believes philosophy can help. He offers a map for navigating rough terrain, from personal trauma to the injustice and absurdity of the world. In this profound and personal book, he shows how the tools of philosophy can help us find our way. Drawing on ancient and modern philosophy as well as fiction, history, memoir, film, comedy, social science, and stories from Setiya’s own experience, “Life Is Hard” is a book for this moment — a work of solace and compassion.

“Grand Delusion: The Rise and Fall of American Ambition in the Middle East” (Penguin, 2023)
By Steven Simon, Robert E. Wilhelm Fellow in the Center for International Studies

The culmination of almost 40 years at the highest levels of policymaking and scholarship, “Grand Delusion” offers a comprehensive and deeply informed account of U.S. engagement in the Middle East. This story, while episodically impressive, was too often tragic and at times dishonorable. As we enter a new era in foreign policy, this is an essential book, a cautionary history that illuminates American’s propensity for self-deception and misadventure at a moment when the nation is redefining its engagement with a world in crisis.

“Ruderal City: Ecologies of Migration, Race, and Urban Nature in Berlin” (Duke University Press, 2022)
By Bettina Stoetzer, associate professor of anthropology

In “Ruderal City,” Stoetzer traces relationships among people, plants, and animals in contemporary Berlin as they make their lives in the ruins of European nationalism and capitalism. She develops the notion of the ruderal — originally an ecological designation for the unruly life that inhabits inhospitable environments such as rubble, roadsides, train tracks, and sidewalk cracks — to theorize Berlin as a “ruderal city.”

“Judicial Dispute Resolution: New Roles for Judges in Ensuring Justice” (Anthem Press, 2023)
By Larry Susskind, the Ford Professor of Urban and Environmental Planning; Justice Williams Tillman; and Nicolas Parra Herrera

Concerned about the role of the courts, particularly judges, in guaranteeing justice, and impressed with the success of Canadian courts that are using judicial dispute resolution (JDR), the authors describe similar efforts in other parts of the world where the use of JDR helps parties resolve their differences in a timely way. The judges who use this practice mediate rather than adjudicate; they do not decide who is right or wrong but assist the parties in resolving their differences and mending their relationships. The authors can tell this unique story after being granted exclusive access to the parties, judges, and records in nine carefully selected cases. 

“Harvard Square: A Love Story” (Columbia University Press, 2023)
By Catherine J. Turco, associate professor of management

In this book, Turco explores the history, impact, and future of street-level markets through her own experience visiting the iconic Cambridge, Massachusetts, neighborhood as a young girl, living there as a university student, and later advocating for the community as a resident.

Technology and society

“Power and Progress: Our Thousand-Year Struggle Over Technology and Prosperity” (PublicAffairs, 2023)
By Daron Acemoglu, Institute Professor, and Simon Johnson, the Ronald A. Kurtz Professor of Entrepreneurship

In their new book, Acemoglu and Johnson detail the ways that artificial intelligence and other digital technologies have mesmerized the business elite while threatening to undermine jobs and democracy. The authors decry the economic and social damage caused by excessive automation, massive data collection, and intrusive surveillance, and offer a counter vision whereby the tremendous computing advances of the past half century can become empowering and democratizing tools.

“Handbook of Innovation and Appropriate Technologies for International Development” (Edward Elgar Publishing, 2022)
Edited by Daniel Frey, professor of mechanical engineering; Samuel Pierre; Philippe Régnier; Koshy Varghese; and Pascal Wild

This book highlights innovations and appropriate technologies helpful for the development of people around the world and across economic backgrounds. An illuminating and informative look into more sustainable global technological development, the handbook presents new disruptive forms of innovation-producing technologies and origin stories about pathbreaking practitioners and organizations. And it provides both the traditional socioeconomic and political frameworks for appropriate technologies and alternative solutions for sustainable development.

“Handbook of Space Resources” (Springer, 2023)
Chapter by Olivier de Weck and Jeffrey Hoffman, professors of aeronautics and astronautics, and George Lordos MBA ’00, SM ’18

This book covers the latest understandings of space resources, including mission concepts, exploration approaches, mining and extraction technologies, commercial potential, and regulation. In their chapter, “Lifetime Embodied Energy: A Theory of Value for the New Space Economy,” adapted from Lordos’ MIT master’s thesis, the authors advance a method based on embodied energy to objectively value space systems that utilize space resources.

“Atlas of the Senseable City” (Yale University Press, 2023)
By Carlo Ratti, professor of the practice and director of the Senseable City Lab, and Antoine Picon

This book explores how the growth of digital mapping, spurred by sensing technologies, is affecting cities and daily lives. It examines how new cartographic possibilities aid urban planners, technicians, politicians, and administrators; how digitally mapped cities could reveal ways to make cities smarter and more efficient; how monitoring urbanites has political and social repercussions; and how the proliferation of open-source maps and collaborative platforms can aid activists and vulnerable populations.

“Reclaiming Space” (Oxford University Press, 2023)
Chapter by Danielle Wood, assistant professor of media arts and sciences, and of aeronautics and astronautics

“Reclaiming Space” is an innovative study of space travel’s history, legitimacy, and future. Wood’s chapter, “Opportunities to Pursue Liberatory, Anticolonial, and Antiracist Designs for Human Societies Beyond Earth,” is one of 27 original essays seeking to incubate, illuminate, and illustrate a more diverse and inclusive conversation about space exploration.

Work, finance, and industry

“Fiscal Policy under Low Interest Rates” (MIT Press, 2023)
By Olivier Blanchard, the Robert Solow Professor of Economics Emeritus

Policy makers in advanced economies find themselves in an unusual fiscal environment: debt ratios are historically high, and — once the fight against inflation is won — real interest rates will likely be very low again. This combination calls for a rethinking of the role of fiscal and monetary policy — and this is just what Blanchard proposes in this work. His conclusions hold practical implications for economic and fiscal policymakers, bankers, and politicians around the world.

“Risky Business: Why Insurance Markets Fail and What to Do About It” (Yale University Press, 2022)
By Amy Finkelstein, professor of economics; Liran Einav; and Ray Fisman

Why is dental insurance so crummy? Why is pet insurance so expensive? Why does your auto insurer ask for your credit score? The answer to these questions lies in understanding how insurance works. Unraveling the mysteries of insurance markets, the authors explore such issues as why insurers want to know so much about us and whether we should let them obtain this information; why insurance entrepreneurs often fail (and some tricks that may help them succeed); and whether we’d be better off with government-mandated health insurance instead of letting businesses, customers, and markets decide who gets coverage and at what price.

“Workforce Ecosystems: Reaching Strategic Goals with People, Partners, and Technologies” (MIT Press, 2023)
By David Kiron, editorial director of research at MIT Sloan Management Review; Elizabeth J. Altman; Jeff Schwartz; and Robin Jones

“Workforce Ecosystems” is a research-driven framework for leading complex, interconnected workforces. Drawing on case studies, worldwide surveys, and extensive interviews with C-suite executives and senior leaders from Amazon, IBM, Mayo Clinic, NASA, Nike, Roche, Unilever, the US Army, Walmart, and others, the authors explore what workforce ecosystems are and how to navigate their unique challenges and opportunities.

“What is This Management?: Essays on Corporate Governance and Management Education” (Self-published, 2023)
By William F. Pounds, professor emeritus of management

In this collection of short essays on management and corporate governance, Pounds shares practical, down-to-Earth wisdom and insight on topics rarely touched on in the typical MBA curriculum, gleaned from his decades of experience across a wide variety of boards. CEOs, corporate directors, and anyone interested in how organizations function and perform in the mysterious realm beyond the executive suite will find these timeless ideas a thought-provoking and sometimes irreverent complement to more traditional academic and legal treatments of these important subjects. 

“The Case for Good Jobs: How Great Companies Bring Dignity, Pay, and Meaning to Everyone’s Work” (Harvard Business Review Press, 2023)
By Zeynep Ton, professor of the practice in operations management

This book serves as a leadership guide for choosing excellence and providing good jobs that offer a living wage, dignity, and opportunities for growth. From health care facilities to call centers, fulfillment centers to factories, and restaurants to retail stores, companies are struggling to find or keep workers, because the jobs they offer are low-paying, stressful, and provide little chance for growth and success. Ton outlines the importance of investing in employees, and the four operational choices managers must make if they want to prioritize customers and maximize employees’ productivity, motivation, and contributions.

“The Magic Conveyer Belt: Supply Chains, A.I., and the Future of Work” (MIT CTL Media, 2023)
By Yossi Sheffi, the Elisha Grey II Professor of Engineering Systems and director of the MIT Center for Transportation and Logistics

In this book, Sheffi takes a close look at “the underlying structure, unavoidable complexity, and massive scale of modern supply chains.” He also explores how automation, robotics, and artificial intelligence are changing and augmenting jobs held by workers and how they will change supply chains of the future.

“Future Ready: The Four Pathways to Capturing Digital Value” (Harvard Business Review, 2022)
By Stephanie L. Woerner, Center for Information Systems Research (CISR) director and principal research scientist; Peter Weill, CISR senior research scientist and chairman emeritus, and Ina M. Sebastian, CISR research scientist 

Companies that undergo digital transformation have significantly higher financial performance. In “Future Ready,” the authors offer a playbook for leaders who want to help their companies leverage digital capabilities to innovate, satisfy customers, and reduce costs. The book provides board members and top management teams leading a digital transformation journey with a coherent framework and a common language to guide, motivate, and focus employees. It is based on more than 50 interviews with executives and surveys with over 2,000 respondents, and was field-tested in multiple workshops with boards and senior management teams in firms around the world.

Arts, architecture, planning, and design

“Equity, Evaluation, and International Cooperation: In Pursuit of Proximate Peers in an African City” (Oxford University Press, 2022)
By Gabriella Carolini, associate professor of urban studies and planning

In this book, Carolini emphasizes that equitable partnership on the ground delivers the best results in the Global South. In her view, the best development projects involve close cooperation between proximate peers: the sharing of information among partners, a consistent presence on the ground, nonhierarchical governance, and a drive toward “equity,” in many forms, as a key project goal.

“Climate Inheritance” (Actar Publishers, 2023)
By Rania Ghosn, associate professor of architecture, and El Hadi Jazairy

The speculative design research publication reckons with the complexity of “heritage” and “world” in the Anthropocene. The impacts of climate change on heritage sites — from Venice flooding to extinction in the Galápagos Islands — have garnered empathetic attention in a media landscape that has otherwise mostly failed to communicate the urgency of the climate crisis. In a strategic subversion of the media aura of heritage, World Heritage sites are cast as narrative figures to visualize pervasive climate risks all while situating the present emergency within the wreckages of other ends of worlds, replete with the salvages of extractivism, racism, and settler colonialism. The harms and possibilities of such inheritances are narrated in drawing triptychs and mythologies to bequeath other worlds and values. 

“Architecture Constructed: Notes on a Discipline” (Bloomsbury Visual Arts, 2023)
By Mark Jarzombek, professor of architecture

This book argues that the architecture/contractor divide is a “construction” with a particular history and theoretical problematic that impacts not just the history of the discipline, but also the history of labor that haunts the very understanding of contemporary architecture. The book looks at issues relating to preservation theory and the Library of Congress ordering systems, as well as to the tragic dualisms of “theory and practice,” mind and body, design and craft, and architect and builder that belay any attempt to ever get out from under its particular falsifications.

“Design and Solidarity: Conversations on Collective Futures” (Columbia University Press, 2023)
By Rafi Segal, associate professor of architecture, and Marissa Morán Jahn SM ’07

The authors converse about the transformative potential of mutualism and design with leading thinkers and practitioners. Together, they consider how design inspires, invigorates, and sustains contemporary forms of mutualism — including platform cooperatives, digital-first communities, emerging currencies, mutual aid, care networks, social-change movements, and more. From these dialogues emerge powerful visions of futures guided by communal self-determination and collective well-being.

“Dare to Know: Prints and Drawings in the Age of Enlightenment” (Yale University Press, 2022)
Edited by Kristel Smentek, associate professor of architecture; Edouard Kopp; and Elizabeth M. Rudy

Are volcanoes punishment from God? During the Enlightenment, questions such as this were brought to life through an astonishing array of prints and drawings, helping shape public opinion and stir political change. This book overturns common assumptions about the Age of Enlightenment, using the era’s proliferation of works on paper to tell a more nuanced story. With a multidisciplinary approach, the book probes developments in the natural sciences, technology, economics, and more — all through the lens of the graphic arts.

“[SpaINdia] Cooper ACTioning Urban Condensers Updating Heritage Areas” (Universidad Politécnica de Madrid Press, 2022)
By Franca Alexandra Sonntag, postdoc and lecturer in architecture, and Ricardo Montoro Coso

In 2019 two Siamese workshops were organized by Aula Coopera [Spain/in/India], under the same research topic, Urban Condensers. Updating heritage areas, the workshops took place in two different cities: Madrid and Ahmedabad; two countries in two continents. With the participation of 14 professors, eight conferences, previous trips visiting paradigmatic architectures, walking through the intervention sites in contact with their inhabitants; 50 students of various levels and seven nationalities produced different proposals for the re-activation of public space in both cities among 17 working groups. This book describes the workshops, including the immense experiences of participants living in situ together.

“Shapes of Imagination: Calculating in Coleridge’s Magical Realm” (MIT Press, 2022)
By George Stiny, professor of architecture

In this book, Stiny runs visual calculating in shape grammars through art and design — incorporating Samuel Taylor Coleridge's poetic imagination and Oscar Wilde's corollary to see things as they aren't. Many assume that calculating limits art and design to suit computers, but shape grammars rely on seeing to prove otherwise. Rules that change what they see extend calculating to overtake what computers can do, in logic and with data and learning. Shape grammars bridge the divide between seeing and combinatoric play.

For young readers

“The Order of Things” (Nancy Paulsen Books, 2023)
By Kaija Langley, director of development at MIT Libraries

“The Order of Things” is a heart-rending and layered novel-in-verse for middle-grade readers. Eleven-year-old April Jackson loves playing the drums, almost as much as she loves her best friend, Zee, a violin prodigy. They both dream of becoming professional musicians one day. When the unthinkable happens and Zee suddenly passes away, April is crushed by grief but begins to learn it is possible to go on even after a great loss.

“I Was Left Out” (Self-published, 2023)
By Sally O. Lee, senior administrative assistant at the Computer Science and Artificial Intelligence Laboratory

Who hasn’t been left out at one point or another? Were you ever not invited to a party, or left out of an occasion? What do you do about it? How do you feel? Who do you talk to? This book follows a kitty as he navigates through his feelings and thoughts and decides what to do next.

de MIT News https://ift.tt/L5BWITe

Computer vision system marries image recognition and generation

Computers possess two remarkable capabilities with respect to images: They can both identify them and generate them anew. Historically, these functions have stood separate, akin to the disparate acts of a chef who is good at creating dishes (generation), and a connoisseur who is good at tasting dishes (recognition).

Yet, one can’t help but wonder: What would it take to orchestrate a harmonious union between these two distinctive capacities? Both chef and connoisseur share a common understanding in the taste of the food. Similarly, a unified vision system requires a deep understanding of the visual world.

Now, researchers in MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) have trained a system to infer the missing parts of an image, a task that requires deep comprehension of the image's content. In successfully filling in the blanks, the system, known as the Masked Generative Encoder (MAGE), achieves two goals at the same time: accurately identifying images and creating new ones with striking resemblance to reality. 

This dual-purpose system enables myriad potential applications, like object identification and classification within images, swift learning from minimal examples, the creation of images under specific conditions like text or class, and enhancing existing images.

Unlike other techniques, MAGE doesn't work with raw pixels. Instead, it converts images into what’s called “semantic tokens,” which are compact, yet abstracted, versions of an image section. Think of these tokens as mini jigsaw puzzle pieces, each representing a 16x16 patch of the original image. Just as words form sentences, these tokens create an abstracted version of an image that can be used for complex processing tasks, while preserving the information in the original image. Such a tokenization step can be trained within a self-supervised framework, allowing it to pre-train on large image datasets without labels. 

Now, the magic begins when MAGE uses “masked token modeling.” It randomly hides some of these tokens, creating an incomplete puzzle, and then trains a neural network to fill in the gaps. This way, it learns to both understand the patterns in an image (image recognition) and generate new ones (image generation).

“One remarkable part of MAGE is its variable masking strategy during pre-training, allowing it to train for either task, image generation or recognition, within the same system,” says Tianhong Li, a PhD student in electrical engineering and computer science at MIT, a CSAIL affiliate, and the lead author on a paper about the research. “MAGE's ability to work in the ‘token space’ rather than ‘pixel space’ results in clear, detailed, and high-quality image generation, as well as semantically rich image representations. This could hopefully pave the way for advanced and integrated computer vision models.” 

Apart from its ability to generate realistic images from scratch, MAGE also allows for conditional image generation. Users can specify certain criteria for the images they want MAGE to generate, and the tool will cook up the appropriate image. It’s also capable of image editing tasks, such as removing elements from an image while maintaining a realistic appearance.

Recognition tasks are another strong suit for MAGE. With its ability to pre-train on large unlabeled datasets, it can classify images using only the learned representations. Moreover, it excels at few-shot learning, achieving impressive results on large image datasets like ImageNet with only a handful of labeled examples.

The validation of MAGE's performance has been impressive. On one hand, it set new records in generating new images, outperforming previous models with a significant improvement. On the other hand, MAGE topped in recognition tasks, achieving an 80.9 percent accuracy in linear probing and a 71.9 percent 10-shot accuracy on ImageNet (this means it correctly identified images in 71.9 percent of cases where it had only 10 labeled examples from each class).

Despite its strengths, the research team acknowledges that MAGE is a work in progress. The process of converting images into tokens inevitably leads to some loss of information. They are keen to explore ways to compress images without losing important details in future work. The team also intends to test MAGE on larger datasets. Future exploration might include training MAGE on larger unlabeled datasets, potentially leading to even better performance. 

“It has been a long dream to achieve image generation and image recognition in one single system. MAGE is a groundbreaking research which successfully harnesses the synergy of these two tasks and achieves the state-of-the-art of them in one single system,” says Huisheng Wang, senior staff software engineer of humans and interactions in the Research and Machine Intelligence division at Google, who was not involved in the work. “This innovative system has wide-ranging applications, and has the potential to inspire many future works in the field of computer vision.” 

Li wrote the paper along with Dina Katabi, the Thuan and Nicole Pham Professor in the MIT Department of Electrical Engineering and Computer Science and a CSAIL principal investigator; Huiwen Chang, a senior research scientist at Google; Shlok Kumar Mishra, a University of Maryland PhD student and Google Research intern; Han Zhang, a senior research scientist at Google; and Dilip Krishnan, a staff research scientist at Google. Computational resources were provided by Google Cloud Platform and the MIT-IBM Watson Research Collaboration. The team's research was presented at the 2023 Conference on Computer Vision and Pattern Recognition.

de MIT News https://ift.tt/ciA91ah

Researchers uncover a new CRISPR-like system in animals that can edit the human genome

A team of researchers led by Feng Zhang at the McGovern Institute for Brain Research at MIT and the Broad Institute of MIT and Harvard has uncovered the first programmable RNA-guided system in eukaryotes — organisms that include fungi, plants, and animals.

In a study published today in Nature, the team describes how the system is based on a protein called Fanzor. They showed that Fanzor proteins use RNA as a guide to target DNA precisely, and that Fanzors can be reprogrammed to edit the genome of human cells. The compact Fanzor systems have the potential to be more easily delivered to cells and tissues as therapeutics than CRISPR-Cas systems, and further refinements to improve their targeting efficiency could make them a valuable new technology for human genome editing.

CRISPR-Cas was first discovered in prokaryotes (bacteria and other single-cell organisms that lack nuclei) and scientists including those in Zhang’s lab have long wondered whether similar systems exist in eukaryotes. The new study demonstrates that RNA-guided DNA-cutting mechanisms are present across all kingdoms of life.

“CRISPR-based systems are widely used and powerful because they can be easily reprogrammed to target different sites in the genome,” says Zhang, senior author on the study, the James and Patricia Poitras Professor of Neuroscience in the MIT departments of Biological Engineering and Brain and Cognitive Sciences, an investigator at MIT’s McGovern Institute, a core institute member at the Broad Institute, and a Howard Hughes Medical Institute investigator. “This new system is another way to make precise changes in human cells, complementing the genome editing tools we already have.”

Searching the domains of life

A major aim of the Zhang lab is to develop genetic medicines using systems that can modulate human cells by targeting specific genes and processes. “A number of years ago, we started to ask, ‘What is there beyond CRISPR, and are there other RNA-programmable systems out there in nature?’” says Zhang.

Two years ago, Zhang lab members discovered a class of RNA-programmable systems in prokaryotes called OMEGAs, which are often linked with transposable elements, or “jumping genes,” in bacterial genomes and likely gave rise to CRISPR-Cas systems. That work also highlighted similarities between prokaryotic OMEGA systems and Fanzor proteins in eukaryotes, suggesting that the Fanzor enzymes might also use an RNA-guided mechanism to target and cut DNA.

In the new study, the researchers continued their work on RNA-guided systems by isolating Fanzors from fungi, algae, and amoeba species, in addition to a clam known as the northern quahog. Co-first author Makoto Saito of the Zhang lab led the biochemical characterization of the Fanzor proteins, showing that they are DNA-cutting endonuclease enzymes that use nearby non-coding RNAs known as ωRNAs to target particular sites in the genome. It is the first time this mechanism has been found in eukaryotes, such as animals.

Unlike CRISPR proteins, Fanzor enzymes are encoded in the eukaryotic genome within transposable elements, and the team’s phylogenetic analysis suggests that the Fanzor genes have migrated from bacteria to eukaryotes through so-called horizontal gene transfer.

“These OMEGA systems are more ancestral to CRISPR and they are among the most abundant proteins on the planet, so it makes sense that they have been able to hop back and forth between prokaryotes and eukaryotes,” says Saito.

No collateral damage

To explore Fanzor’s potential as a genome editing tool, the researchers demonstrated that it can generate insertions and deletions at targeted genome sites within human cells. The researchers found the Fanzor system to initially be less efficient at snipping DNA than CRISPR-Cas systems, but by systematic engineering, they introduced a combination of mutations into the protein that increased its activity 10-fold. Additionally, unlike some CRISPR systems and the OMEGA protein TnpB, the team found that a fungal-derived Fanzor protein did not exhibit “collateral activity,” where an RNA-guided enzyme cleaves its DNA target as well as degrading nearby DNA or RNA. The results suggest that Fanzors could potentially be developed as efficient genome editors.

Co-first author Peiyu Xu led an effort to analyze the molecular structure of the Fanzor/ωRNA complex and illustrate how it latches onto DNA to cut it. Fanzor shares structural similarities with its prokaryotic counterpart CRISPR-Cas12 protein, but the interaction between the ωRNA and the catalytic domains of Fanzor is more extensive, suggesting that the ωRNA might play a role in the catalytic reactions. “We are excited about these structural insights for helping us further engineer and optimize Fanzor for improved efficiency and precision as a genome editor,” said Xu.

Like CRISPR-based systems, the Fanzor system can be easily reprogrammed to target specific genome sites, and Zhang said it could one day be developed into a powerful new genome editing technology for research and therapeutic applications. The abundance of RNA-guided endonucleases like Fanzors further expands the number of OMEGA systems known across kingdoms of life and suggests that there are more yet to be found.

“Nature is amazing. There’s so much diversity,” says Zhang. “There are probably more RNA-programmable systems out there, and we’re continuing to explore and will hopefully discover more.”

The paper’s other authors include Guilhem Faure, Samantha Maguire, Soumya Kannan, Han Altae-Tran, Sam Vo, AnAn Desimone, and Rhiannon Macrae.

Support for this work was provided by the Howard Hughes Medical Institute; Poitras Center for Psychiatric Disorders Research at MIT; K. Lisa Yang and Hock E. Tan Molecular Therapeutics Center at MIT; Broad Institute Programmable Therapeutics Gift Donors; The Pershing Square Foundation, William Ackman, and Neri Oxman; James and Patricia Poitras; BT Charitable Foundation; Asness Family Foundation; Kenneth C. Griffin; the Phillips family; David Cheng; Robert Metcalfe; and Hugo Shong.

de MIT News https://ift.tt/YyAZqcv

Gamifying medical data labeling to advance AI

When Erik Duhaime PhD ’19 was working on his thesis in MIT’s Center for Collective Intelligence, he noticed his wife, then a medical student, spending hours studying on apps that offered flash cards and quizzes. His research had shown that, as a group, medical students could classify skin lesions more accurately than professional dermatologists; the trick was to continually measure each student’s performance on cases with known answers, throw out the opinions of people who were bad at the task, and intelligently pool the opinions of people that were good.

Combining his wife’s studying habits with his research, Duhaime founded Centaur Labs, a company that created a mobile app called DiagnosUs to gather the opinions of medical experts on real-world scientific and biomedical data. Through the app, users review anything from images of potentially cancerous skin lesions or audio clips of heart and lung sounds that could indicate a problem. If the users are accurate, Centaur uses their opinions and awards them small cash prizes. Those opinions, in turn, help medical AI companies train and improve their algorithms.

The approach combines the desire of medical experts to hone their skills with the desperate need for well-labeled medical data by companies using AI for biotech, developing pharmaceuticals, or commercializing medical devices.

“I realized my wife’s studying could be productive work for AI developers,” Duhaime recalls. “Today we have tens of thousands of people using our app, and about half are medical students who are blown away that they win money in the process of studying. So, we have this gamified platform where people are competing with each other to train data and winning money if they’re good and improving their skills at the same time — and by doing that they are labeling data for teams building life saving AI.”

Gamifying medical labeling

Duhaime completed his PhD under Thomas Malone, the Patrick J. McGovern Professor of Management and founding director of the Center for Collective Intelligence.

“What interested me was the wisdom of crowds phenomenon,” Duhaime says. “Ask a bunch of people how many jelly beans are in a jar, and the average of everybody’s answer is pretty close. I was interested in how you navigate that problem in a task that requires skill or expertise. Obviously you don’t just want to ask a bunch of random people if you have cancer, but at the same time, we know that second opinions in health care can be extremely valuable. You can think of our platform as a supercharged way of getting a second opinion.”

Duhaime began exploring ways to leverage collective intelligence to improve medical diagnoses. In one experiment, he trained groups of lay people and medical school students that he describes as “semiexperts” to classify skin conditions, finding that by combining the opinions of the highest performers he could outperform professional dermatologists. He also found that by combining algorithms trained to detect skin cancer with the opinions of experts, he could outperform either method on its own.

“The core insight was you do two things,” Duhaime explains. “The first thing is to measure people’s performance — which sounds obvious, but even in the medical domain it isn’t done much. If you ask a dermatologist if they’re good, they say, ‘Yeah of course, I’m a dermatologist.’ They don’t necessarily know how good they are at specific tasks. The second thing is that when you get multiple opinions, you need to identify complementarities between the different people. You need to recognize that expertise is multidimensional, so it’s a little more like putting together the optimal trivia team than it is getting the five people who are all the best at the same thing. For example, one dermatologist might be better at identifying melanoma, whereas another might be better at classifying the severity of psoriasis.”

While still pursuing his PhD, Duhaime founded Centaur and began using MIT’s entrepreneurial ecosystem to further develop the idea. He received funding from MIT’s Sandbox Innovation Fund in 2017 and participated in the delta v startup accelerator run by the Martin Trust Center for MIT Entrepreneurship over the summer of 2018. The experience helped him get into the prestigious Y Combinator accelerator later that year.

The DiagnosUs app, which Duhaime developed with Centaur co-founders Zach Rausnitz and Tom Gellatly, is designed to help users test and improve their skills. Duhaime says about half of users are medical school students and the other half are mostly doctors, nurses, and other medical professionals.

“It’s better than studying for exams, where you might have multiple choice questions,” Duhaime says. “They get to see actual cases and practice.”

Centaur gathers millions of opinions every week from tens of thousands of people around the world. Duhaime says most people earn coffee money, although the person who’s earned the most from the platform is a doctor in eastern Europe who’s made around $10,000.

“People can do it on the couch, they can do it on the T,” Duhaime says. “It doesn’t feel like work — it’s fun.”

The approach stands in sharp contrast to traditional data labeling and AI content moderation, which are typically outsourced to low-resource countries.

Centaur’s approach produces accurate results, too. In a paper with researchers from Brigham and Women’s Hospital, Massachusetts General Hospital (MGH), and Eindhoven University of Technology, Centaur showed its crowdsourced opinions labeled lung ultrasounds as reliably as experts did. Another study with researchers at Memorial Sloan Kettering showed crowdsourced labeling of dermoscopic images was more accurate than that of highly experienced dermatologists. Beyond images, Centaur’s platform also works with video, audio, text from sources like research papers or anonymized conversations between doctors and patients, and waves from electroencephalograms (EEGs) and electrocardiographys (ECGs).

Finding the experts

Centaur has found that the best performers come from surprising places. In 2021, to collect expert opinions on EEG patterns, researchers held a contest through the DiagnosUs app at a conference featuring about 50 epileptologists, each with more than 10 years of experience. The organizers made a custom shirt to give to the contest’s winner, who they assumed would be in attendance at the conference.

But when the results came in, a pair of medical students in Ghana, Jeffery Danquah and Andrews Gyabaah, had beaten everyone in attendance. The highest-ranked conference attendee had come in ninth.

“I started by doing it for the money, but I realized it actually started helping me a lot,” Gyabaah told Centaur’s team later. “There were times in the clinic where I realized that I was doing better than others because of what I learned on the DiagnosUs app.”

As AI continues to change the nature of work, Duhaime believes Centaur Labs will be used as an ongoing check on AI models.

“Right now, we’re helping people train algorithms primarily, but increasingly I think we’ll be used for monitoring algorithms and in conjunction with algorithms, basically serving as the humans in the loop for a range of tasks,” Duhaime says. “You might think of us less as a way to train AI and more as a part of the full life cycle, where we’re providing feedback on models’ outputs or monitoring the model.”

Duhaime sees the work of humans and AI algorithms becoming increasingly integrated and believes Centaur Labs has an important role to play in that future.

“It’s not just train algorithm, deploy algorithm,” Duhaime says. “Instead, there will be these digital assembly lines all throughout the economy, and you need on-demand expert human judgment infused in different places along the value chain.”

de MIT News https://ift.tt/M0KaNvG

Uncovering the invisible pressures shaping language

Sihan Chen's fascination with languages began when he was a teenager in his high school choir in his hometown of Shenzhen, China. "We sang religious and folk songs in different languages like Mandarin, English, Albanian, and Latin, and I would look up how to pronounce words in different languages," he says.

These early experiences planted the seed for his current research interests in cognition and language as a third-year PhD student in MIT's Department of Brain and Cognitive Sciences (BCS). Under the guidance of Professor Edward "Ted" Gibson, Chen studies how different languages have developed into what we see today, with a particular emphasis on environmental and social factors. "I was always interested in the invisible pressures that shape our languages over time," he says.
 
Despite Chen's ongoing interest in language, he started on a different academic path. After completing the demanding entrance exams for Chinese universities, he made a last-minute decision to apply to U.S. colleges. With just three months to prepare for the SAT and language test, he says his resume paled compared to his American peers' extensive extracurricular activities. He took the chance and landed at the University of Miami, where he pursued a major in mechanical engineering.
 
Alongside his engineering studies, Chen dedicated several elective courses to his interest in linguistics. In one of these courses, he stumbled upon a thought-provoking paper, coincidentally written by a student from BCS, exploring how language influences our perception of color. "It fascinated me and made quite an impression," he recalls. Now, he is working on a paper challenging some of those findings.
 
Chen says reading such papers in college helped him discover the perfect match between his intellectual interests and abilities, augmented by the skills he had acquired through his mechanical engineering training. Although the topics addressed in these papers were quite different from what he encountered in his major, the underlying thought process to solve the problems are similar. "You identify the process, break it down into different steps, and think about how you can model each mathematically and implement the model computationally." He decided to reach out to Gibson, who further encouraged him to continue his work.
 
Approaching graduation, Chen faced a tough decision between the more familiar path of biomedical engineering and pursuing his deepening passion for language and cognition. "I think it was clear this is where I wanted to go, but it was a tough decision." He decided to apply.

Chen's current work with Gibson revolves around a central thesis of his lab: Since we often use language to communicate, we have the incentive to use it efficiently. Chen and his colleagues are looking into how different languages convey spatial information with spatial deictic demonstratives — words or phrases such as "here" and "from there" in English. "Efficiency here means communicating spatial information as accurately as possible, with a system involving as few demonstratives as possible," says Chen.

In communication, there is a trade-off between two factors: accuracy and simplicity, as one cannot fully achieve both at the same time. If one wants to pursue accuracy, the resulting spatial deictic system must be very complex. In contrast, if one intends to pursue simplicity, the resulting system won't be able to convey any information. Using tools in information theory and a database of 220 languages from various parts of the world, Chen and his collaborators show that existing spatial deictic systems are very close to the optimal systems predicted by information theory. In addition, they also show that human languages prefer spatial demonstrative systems that have a consistent pattern.

Seeking how societal characteristics might shape language features, Chen and his colleagues analyze data from diverse languages. They hypothesize two types of societies: exoteric societies, marked by mobility, large population size, and a significant number of second language speakers, and esoteric societies, characterized by being more close-knit, having less migration, and fewer second-language speakers. Exoteric languages, such as English, are widely spoken and have more complex syntax but simplified morphology. Conversely, esoteric societies tend to have languages with more complex morphology, such as more noun cases.

"English, for example, rarely has cases, except in pronouns where it distinguishes between nominative and accusative cases with pronouns like 'I' and 'me' and uses the genitive case with 'my,' indicating a relationship, and that's about it. But other, more esoteric languages have up to 15 cases, including cases for objects, direction, and location. We think that this difference in noun cases arises from word learning processes, where second-language learners focus on rules, while native speakers memorize these specific variations."
 
Trying to see what influences the way humans speak, Chen's research also extends to acoustics. A waveform is a visual representation of a sound you might be familiar with from your audio player. Waveforms depict the amplitude of a wave over time, comprising delicate and minuscule oscillations that amalgamate into a broader pattern known as the modulation envelope. "We want to determine the frequency of this broader oscillation, which can also be described as the rhythm of speech," Chen explains. Previous research revealed that in eight languages, this rhythm of speech typically exhibits a frequency peaking at 4 hertz — a rhythm that researchers believe is related to the coordination between speech production and comprehension. "We expanded this research to 94 languages, and we are now analyzing 4 million recordings to determine further the prevalence of this phenomenon across different regions of the globe," says Chen.
 
Reflecting on his research and studies in the last three years, Chen says MIT's Department of Brain and Cognitive Sciences is a highly supportive environment, fostering collaboration among individuals from diverse backgrounds. "There is always someone available to offer assistance and guidance," he says. "it's so cool to see people from different backgrounds with their own expertise coming together collaboratively to uncover fascinating secrets of the human mind."

de MIT News https://ift.tt/4T8SBnd

Exploring the unexpected social questions behind everyday medical devices

At the height of the Covid-19 pandemic, when many hospitals ran out of beds and ventilators, the fingertip pulse oximeter — a $20 neighborhood drugstore purchase — became a primary arbiter of whether a patient was “sick enough” to gain admission to an emergency room.

This spring, surrounded by antique telescope models in a classroom tucked inside the MIT Museum, 10 students bent over a square-shaped seminar table. They were building basic pulse oximeters from low-cost do-it-yourself (DIY) kits of assorted wirelings from TinyCircuits. 

This was just one meeting from MIT class 21A.311 (The Social Lives of Medical Objects), a course taught by Amy Moran-Thomas, associate professor of anthropology and the 2022 winner of the Edgerton Faculty Achievement Award.

Moran-Thomas observed that anthropology’s key frameworks for studying material culture were often missing from spaces of device design and pre-med coursework. She developed the class in 2019 to encourage students to learn from ethnographic approaches, the MIT Program in Science, Technology and Society, and history to explore social inquiries related to health objects.

Zeroing in on a case study concerning a different health material, device, or technology, each session probes what Moran-Thomas refers to as “the social assumptions that get built into objects meant to improve health.”

For the week’s focus on “Politics of Measurement,” students rolled up their sleeves for the lab portion of class, joined by Jose Gomez-Marquez of MakerHealth. He guided the teams through assembling their own oximeter. 

Then the students were encouraged to interact with it alongside two prepackaged models that displayed readings from sealed-off “black boxes.” The open-design TinyCircuits models demystified these inner workings, which Moran-Thomas pointed out is a reminder of all the choices that any black box can contain. Students attached the DIY sensors to a microcontroller board, and played with lines of code as they wondered how small edits might impact the readings of sensors held to their fingertips.

Nicole Seman '23, a recent graduate in mechanical engineering, has Type 1 diabetes and uses an insulin pump. She enrolled in the course because being dependent on a medical device is something she knows all too well. In a paper, she wrote about the experience of switching pumps, drawing on an ethnography about what assumptions of “normal” bodies get encoded in global cochlear implants and their sensory futures.

After tinkering with the DIY oximeter, Seman noted how some students immediately reported wide discrepancies when wearing two prepackaged devices at once. “It’s a diverse class, and so we were able to see firsthand how the pulse ox works differently depending on skin tone.”

A class with timely implication

This observation, perhaps obvious in 2023, received tremendous pushback when Moran-Thomas first published a widely read essay in the Boston Review laying out the evidence for how pulse oximeters can offer biased results for people with darker skin.

In the essay, she explained that oximeters gauge oxygen levels in part by measuring color absorption: blood’s iron-containing hemoglobin is brighter crimson-colored when fully saturated with oxygen, and a cooler purple-red when holding less oxygen. To measure this, a pulse oximeter shines two light wavelengths through the skin — but many device developers had not carefully accounted for the ways light is absorbed differently across various skin tones, often designing and testing it with largely white test groups. Distorted measurements can be amplified by algorithms, and Moran-Thomas warned that consequences of this inherent bias could produce device errors with life-altering implications, such as whether a patient is admitted to the hospital or offered oxygen.

In Moran-Thomas’s classroom, students had the opportunity to open up devices and see for themselves the human choices that “black boxes” can encode. It was an elegantly simple example of how easy it is to assume that numbers are neutral in health care — when, in fact, codes and the way they are interpreted are often rife with subtle bias.

She says, “Something I hoped the students would see from this exercise is that many things that are taken to be raw data are actually processed signals, mediated by design decisions and unequal histories. Social assumptions get materialized in how we build technology and make measures.” Those assumptions about who a device’s end users will be directly impacts whether the design and its function are equitable — or dangerously unequal.

The session ended that day with a visit from Tufts University associate professor of electrical and computer engineering and MIT alumna and recent MLK Visiting Professor Valencia Joyner Koomson ’98, MNG ’99. Working in a professional landscape where only 2 percent of engineers are Black women, Koomson presented her work to build a “smart” pulse oximeter that works for everyone, with an eye toward global accessibility. 

Moran-Thomas says that Koomson’s presentation showed why equitable design is important, and conveyed to students how such work means much more than just diverse testing: it also takes diversity in engineering and among experts.

Device tours

For one of the field trips early in the semester, Moran-Thomas arranged a “devices tour” of an MIT ambulance led by junior Abbie Schipper, who volunteers at MIT Emergency Medical Services as an emergency medical technician. Schipper took students through the vitals equipment, including blood pressure monitors and cuffs, and pulse oximeters.

Schipper’s tour also highlighted how wider consciousness of device questions is growing in the medical community. Already well-aware that pulse oximeters do not work equally for everyone, she brought additional examples to the students’ attention: Blood pressure cuffs, which work manually, are notoriously unreliable as the ambulance accelerates down potholed Cambridge roads. “You treat the patient, not the number,” Schipper told the students. “The ambulance itself is a big medical device.”

Students in the class appreciated hearing how Schipper’s own experiences working in emergency transport impacted her thinking about design questions as a mechanical engineering student.

Chloe McCreery, a senior planning to attend medical school, says “This class has really made me think; we all have a responsibility to make sure these devices are created and used ethically.”

Drawing future lessons from historical objects at the MIT Museum

The class also pulled from the Technology Collection at the MIT Museum. At one recent session, students examined one of the museum’s most recent acquisitions, a retro detector known as the Bedside Arrhythmia Monitor. It was rediscovered by someone cleaning out an old building on campus and made its way back to the museum last fall.

“The MIT Museum started to research it, and realized it was a milestone device,” Moran-Thomas says. “They invited our class to contribute to exploring the social side of the story.” Museum curators had been able to ascertain that the rectangular device was prototyped in MIT’s Biomedical Engineering Center for Clinical Instrumentation in the 1970s, and bore metal tags from years it spent in service at Beth Israel Deaconess Medical Center and NASA. 

The Bedside Arrhythmia Monitor helped lead to the creation of PhysioNet, today one of the most widely used open health databases in the world. Moran-Thomas tracked down Paul Scott Schluter, who developed the detector for his thesis, and his advisor, MIT Professor Roger Mark. They spoke with the class virtually to share recollections of how the arrhythmia detector and its pivotal open database came into being. During their chat, the device sat on a table between the generations, and a conversation began about its potential implications for the next chapters of open data design in health.

“Each case study is meant to emphasize one of anthropology’s core lessons: At every stage of an object’s creation and use, there are people with histories and insights to consider,” Moran-Thomas says.

Now, it’s up to her students to bring ethnographic questions and social approaches into their future work.

de MIT News https://ift.tt/Expim0P

Sensing the world around us

“Sensing is all around you,” said MIT.nano Associate Director Brian W. Anthony at Ambient Sensing, a half-day symposium presented in May by the MIT.nano Immersion Lab. Featuring MIT faculty and researchers from multiple disciplines, the event highlighted sensing technologies deployed everywhere from beneath the Earth’s surface to high into the exosphere.

Brent Minchew, assistant professor in the Department of Earth, Atmospheric and Planetary Sciences (EAPS), kicked off the symposium with a presentation on using remote sensing to understand the flow, deformation, and fracture of glacier ice, and how that is contributing to sea level rise. “There’s this fantastic separation of scales,” said Minchew. “We’re taking observations collected from satellites that are flying 700 kilometers above the surface, and we’re using the data that’s collected there to infer what’s happening at the atomic scale within the ice, which is magnificent.”

Minchew’s group is working with other researchers at MIT to build a drone capable of flying for three to four months over the polar regions, filling critical gaps in earth observations. “It’s going to give us this radical improvement over current technology and our observational capacity.”

Also using satellites, EAPS postdoc Qindan Zhu combines machine learning with observational inputs from remote sensing to study ozone pollution over North American cities. Zhu explained that, based on a decade worth of data, controlling nitrogen oxides emissions will be the most effective way to regulate ozone pollution in these urban areas. Both Zhu’s and Minchew’s presentations highlighted the important role ambient sensors play in learning more about Earth’s changing climate.

Transitioning from air to sea, Michael Benjamin, principal research scientist in the Department of Mechanical Engineering, spoke about his work on robotic marine vehicles to explore and monitor the ocean and coastal marine environments. “Robotic platforms as remote sensors have the ability to sense in places that are too dangerous, boring, or costly for crewed vessels,” explained Benjamin. At the MIT Marine Autonomy Lab, researchers are designing underwater surface robots, autonomous sailing vessels, and an amphibious surf zone robot.

Sensing is a huge part of marine robotics, said Benjamin. “Without sensors, robots wouldn’t be able to know where they are, they couldn’t avoid hidden things, they couldn’t collect information.”

Fadel Adib, associate professor​ in the Program in Media Arts & Sciences and the Department of Electrical Engineering & Computer Science (EECS), is also working on sensing underwater. “Battery life of underwater sensors is extremely limited,” explained Adib. “It is very difficult to recharge the battery of an ocean sensor once it’s been deployed.”

His research group built an underwater sensor that reflects acoustic signals rather than needing to generate its own, requiring much less power. They also developed a battery-free, wireless underwater camera that can capture images continuously and over a long period of time. Adib spoke about potential applications for underwater ambient sensing — climate studies, discovery of new ocean species, monitoring aquaculture farms to support food security, and even beyond the ocean, in outer space. “As you can imagine, it’s even more difficult to replace a sensor’s battery once you’ve shipped it on a space mission,” he said.

Originally working in the underwater sensing world, James Kinsey, CEO of Humatics, is applying his knowledge of ocean sensors to two different markets: public transit and automotive manufacturing. “All of that sensor data in the ocean — the value is when you can geolocate it,” explained Kinsey. “The more precisely and accurately you know that, you can begin to paint that 3D space.” Kinsey spoke about automating vehicle assembly lines with millimeter precision, allowing for the use of robotic arms. For subway trains, he highlighted the benefits of sensing systems to better know a train’s position, as well as to improve rider and worker safety by increasing situational awareness. “Precise positioning transforms the world,” he said.

At the intersection of electrical engineering, communications, and imaging, EECS Associate Professor Ruonan Han introduced his research on sensing through semiconductor chips that operate at terahertz frequencies. Using these terahertz chips, Han’s research group has demonstrated high-angular-resolution 3D imaging without mechanical scanning. They’re working on electronic nodes for gas sensing, precision timing, and miniaturizing tags and sensors.

In two Q&A panels led by Anthony, the presenters discussed how sensing technologies interface with the world, highlighting challenges in hardware design, manufacturing, packaging, reducing cost, and producing at scale. On the topic of data visualization, they agreed on a need for hardware and software technologies to interact with and assimilate data in faster, more immersive ways.

Ambient Sensing was broadcast live from the MIT.nano Immersion Lab. This unique research space, located on the third floor of MIT.nano, provides an environment to connect the physical to the digital — visualizing data, prototyping advanced tools for augmented and virtual reality (AR/VR), and developing new software and hardware concepts for immersive experiences.

To showcase current work being done in the Immersion Lab, retired MIT fencing coach Robert Hupp joined Anthony and research scientist Praneeth Namburi for a live demonstration of immersive athlete-training technology. Using wireless sensors on the fencing épée paired with OptiTrack motion-capture sensors along the room’s perimeter, a novice fencer wearing a motion-capture suit and an AR headset faced a virtual opponent while Namburi tracked the fencer’s stance on a computer. Hupp was able to show the fencer how to improve his movements with this real-time data.

“This event showcased the capabilities of the Immersion Lab, and the work being done on sensing — including sensors, data analytics, and data visualization — across MIT,” says Anthony. “Many of our speakers talked about collaboration and the importance of bringing multiple fields together to advance ambient sensing and data collection to solve societal challenges. I look forward to welcome more academic and industry researchers into the Immersion Lab to support their work with our advanced hardware and software technologies.”

de MIT News https://ift.tt/uhoHPyn

A novel combination therapy counters antibiotic-resistant Mycobacterium abscessus infections

Researchers from the Antimicrobial Resistance (AMR) interdisciplinary research group at Singapore-MIT Alliance for Research and Technology (SMART), MIT’s research enterprise in Singapore, in collaboration with Nanyang Technological University Singapore (NTU Singapore) and National University Hospital, have discovered a novel therapy by combining two antibiotics, rifaximin and clarithromycin, to treat Mycobacterium abscessus, a non-tuberculous mycobacterium (NTM) that causes chronic lung-related infections.

Infections caused by NTM are a fast-growing health concern worldwide, particularly in the context of lung-related infections. Among NTMs, M. abscessus is one of the most prevalent, causing pulmonary infections in humans with immune deficiencies or underlying lung conditions. M. abscessus has also been linked to severe infections in various other parts of the body, including the skin, joints, soft tissues, and surgical sites. These infections are difficult to treat due to the bacterium’s extensive innate resistance to many commonly used antibiotics.

Currently, M. abscessus infections are treated by a multidrug regimen that includes clarithromycin, but some M. abscessus subspecies acquire resistance upon repeated exposure to the drug. As a result, available treatment options are limited, leading to prolonged and recurrent infections and even fatalities in some cases. With clarithromycin being the mainstay of NTM treatments and currently the only highly effective oral antibiotic for treating M. abscessus infections, there is an urgent medical need for the identification of compounds that are clarithromycin potentiators in order to effectively restore clarithromycin efficacy against M. abscessus.

A recent open-access study by SMART researchers, “Rifaximin potentiates clarithromycin against Mycobacterium abscessus in vitro and in zebrafish,” published in the scientific journal JAC-Antimicrobial Resistance, revealed promising findings on the use of rifaximin, an antibiotic commonly used to treat gastrointestinal bacterial infections, as a clarithromycin potentiator with the ability to increase clarithromycin sensitivity and improve its ability to kill M. abscessus. During the discovery stage of the study, the researchers conducted drug screening campaigns and successfully identified several drug candidates as clarithromycin potentiators. Further preclinical testing of these drug candidates confirmed rifaximin as the most effective clarithromycin potentiator, with the combination of rifaximin and clarithromycin showing efficacy both in vitro and in a zebrafish embryo infection model.

“With limited treatments available due to M. abscessus's innate resistance to most antibiotics, including clarithromycin, the novel discovery of the strong combination between rifaximin and clarithromycin is a significant step towards addressing the challenge of treating NTM infections. As FDA-approved drugs, we will be able to quicken the process and translate the findings into improved treatment outcomes for patients suffering from M. abscessus infections,” says MIT professor of biological engineering Peter C. Dedon, senior author of the paper and co-lead principal investigator at SMART AMR.

“We recognize the urgent need to address the growing problem of clarithromycin resistance in M. abscessus and are pleased to have discovered rifaximin as a potent clarithromycin potentiator. Despite being primarily used for gastrointestinal infections and having limited activity against drug-resistant M. abscessus, our study demonstrated the synergistic effect of rifaximin with clarithromycin in effectively eliminating the bacteria,” adds Boon Chong Goh, first author of the paper and principal research scientist at SMART AMR.

The researchers are furthering their research with animal preclinical studies to prepare for human clinical trials. As both rifaximin and clarithromycin are approved by the U.S. Food and Drug Administration, the preclinical studies evaluating their combination against M. abscessus can be accelerated. The team is also collaborating with a commercial manufacturing partner to create inhalation formulations suitable for delivering the drug combination directly to the lungs for use in human clinical trials.

The research was carried out by SMART and supported by the A*STAR Singapore Therapeutics Development Review, the SMART Innovation Centre, and the National Research Foundation (NRF) Singapore under its Campus for Research Excellence And Technological Enterprise (CREATE) program. The NTU Singapore researchers played an important role in performing the zebrafish embryo infection model, and NUH provided the clinical isolates of M. abscessus.

The AMR IRG is a translational research and entrepreneurship program that tackles the growing threat of antimicrobial resistance. By leveraging talent and convergent technologies across Singapore and MIT, they tackle AMR head-on by developing multiple innovative and disruptive approaches to identify, respond to, and treat drug-resistant microbial infections. Through strong scientific and clinical collaborations, they provide transformative, holistic solutions for Singapore and the world. SMART was established by MIT and the NRF is 2007. SMART is the first entity in CREATE, undertaking cutting-edge research in areas of interest to both Singapore and MIT.

de MIT News https://ift.tt/24mDLIz