Massachusetts Institute of Technology faculty

"We reject: kings, presidents and voting. We believe in: rough consensus and running code."

"If history has a lesson, it is that the "winner take all" attitude deprives one of the pleasures of being the heir to the best of different traditions, even while avoiding their intolerance against each other."

"There is no concept in the whole field of physics which is more difficult to understand than is the concept of entropy, nor is there one which is more fundamental."

"A patent is a legal analog of sticky fly paper: it attracts some of the lowest forms of life."

"There is an ancient Chinese saying "He who labours with his mind rules over he who labours with his hand". This kind of backward idea is very harmful to youngsters from developing countries. Partly because of this type of concept, many students from these countries are inclined towards theoretical studies and avoid experimental work. In reality, a theory in natural science cannot be without experimental foundations; physics, in particular, comes from experimental work."

"Archaeology of the future is what it should be called. Archaeology of the past is very interesting because it tells us what we once were. But archaeology of the future is the study of what we're going to become, what we have a chance to become...it's a missing element in our understanding of the universe which tells us what our future is like, and what our place in the universe is. If there's nobody else out there, that's also quite important to know."

"There is a narrowness of action, though not of intent, which characterizes university departments, and scientific publications and scientists in general: if it is too popular, it is somehow vulgar and wrong. You can't really speak to those people across the street. I live next to the chemists at MIT, but I never see them. I hardly know who they are, yet between physics and chemistry it is hard to know who should study what molecule. I myself am guilty. We form communities not based on the problems of science, but on quite other things. This is part of the general split between the intelligent member of the public and the scientist who speaks in narrow focus. But the great theoretical problems which I believe the world expects will somehow be solved by science, problems close to deep philosophical issues are the very problems that find the least expertise, the least degree of organization, the least institutional support in the scientific institutions of America or indeed of the world."

"There has never been just one best way to teach quantum mechanics. My goal is neither to sow nostalgia for the philosophically engaged style of Oppenheimer and Nordheim, nor to condemn the pragmatic approach of Fermi, Bethe and Feynman. It is rather to highlight the choices that physicists must always make when stepping into the classroom. Choices of topics to discuss and problems to assign reflect deeper decisions about the ideal type of physicist one seeks to train. Should the new generation be philosophically attuned, concerned with minute details of conceptual interpretation? Or should physicists hone their ability to calculate, pushing Heisenberg’s and Schrödinger’s equations into the service of ever more elaborate problems to solve and phenomena to analyse? Competing ideals have flourished under different pedagogical conditions."

"Strangely enough, many of the philosophical issues surrounding quantum mechanics are today being used to entice potential students into physics. As quantum computing and quantum communication become a commercial reality, tomorrow’s students may find themselves routinely grappling with the same philosophical questions that challenged their forebears almost a century ago."

"Management cannot provide a man with self-respect or with the respect of his fellows or with the satisfaction of needs for self-fulfillment. It can create conditions such that he is encouraged and enabled to seek such satisfactions for himself, or it can thwart him by failing to create those conditions."

"The key question for top management is what are your assumptions (implicit as well as explicit) about the most effective way to manage people?"

"Every managerial act rests on assumptions, generalizations, and hypotheses — that is to say, on theory. Our assumptions are frequently implicit, sometimes quite unconscious, often conflicting; nevertheless, they determine our predictions that if we do a, b will occur. Theory and practice are inseparable."

"Human behavior is predictable, but, as in physical science, accurate prediction hinges on the correctness of underlying theoretical assumptions. There is, in fact, no prediction without theory; all managerial decisions and actions rest on assumptions about behavior. If we adopt the posture of the ostrich with respect to our assumptions under the mistaken idea that we are thus “being ‘practical,” or that “management is an art,” our progress with respect to the human side of enterprise will indeed be slow. Only as we examine and test our theoretical assumptions can we hope to make them more adequate, to remove inconsistencies, and thus to improve our ability to predict."

"The ingenuity of the average worker is sufficient to outwit any system of controls devised by management."

"Formal theories of organization have been taught in management courses for many years, and there is an extensive literature on the subject. The textbook principles of organization — hierarchical structure, authority, unity of command, task specialization, division of staff and line, span of control, equality of responsibility and authority, etc. — comprise a logically persuasive set of assumptions which have had a profound influence upon managerial behavior."

"Classical organization theory suffers from "ethnocentrism": It ignores the significance of the political, social, and economic milieu in shaping organizations and influencing managerial practice."

"We live today in a world which only faintly resembles that of a half century ago. The standard of living, the level of education, and the political complexion of the United States today profoundly affect both the possibilities and limitations of organizational behavior. In addition, technological changes are bringing about changes in all types of organization. In the military, for example, it is becoming increasingly difficult to manage a weapons team in the field as a typical infantry unit was managed a couple of decades ago. Such a team requires a high degree of autonomy. Instead of following explicit orders from superiors, it must be able to adjust its behavior to fit local circumstances within the context of relatively broad objectives. (It is interesting to note the attempts that are made — by "programming" for example — to retain central control over the operations of such units. Established theories of control are not abandoned easily, even in the face of clear evidence of their inappropriateness.) Underlying the principles of classical organization theory are a number of assumptions about human behavior which are at best only partially true."

"Knowledge accumulated during recent decades challenges and contradicts assumptions which are still axiomatic in conventional organizational theory. Unfortunately, those classical principles of organization — derived from inappropriate models, unrelated to the political, social, economic, and technological milieu, and based on erroneous assumptions about behavior — continue to influence our thinking about the management of the human resources of industry. Management's attempts to solve the problems arising from the inadequacy of these assumptions have often involved the search for new formulas, new techniques, new procedures. These generally yield disappointing results because they are adjustments to symptoms rather than causes. The real need is for new theory, changed assumptions, more understanding of the nature of human behavior in organizational settings."

"If there is a single assumption which pervades conventional organizational theory, it is that authority is the central, indispensable means of managerial control."

"The effectiveness of authority as a means of control depends first of all upon the ability to enforce it through the use of punishment. In the two organizations which have been the models for classical organization theory, the situation with respect to enforcement is clear. In the military, authority is enforceable through the court-martial, with the death penalty as the extreme form of punishment. In the Church, excommunication represents the psychological equivalent of the death penalty."

"Behind every managerial decision or action are assumptions about human nature and human behavior."

"The average human being has an inherent dislike of work and will avoid it if he can."

"The capacity to exercise a relatively high degree of imagination, ingenuity and creativity in the solution of organizational problems is widely, not narrowly, distributed in the population."

"It is probable that one day we shall begin to draw organization charts as a series of linked groups rather than as a hierarchical structure of individual "reporting" relationships."

"The essential task of management is to arrange organizational conditions and methods of operations so that people can achieve their own goals best by directing their own efforts toward organizational objectives."

"Above all, it is necessary to recognize that knowledge cannot be pumped into human beings the way grease is forced into a machine. The individual may learn; he is not taught."

"Delegation means that he will concern himself with the results of their activities and not with the details of their day-to-day performance. This requires a degree of confidence in them which enables him to accept certain risks. Unless he takes these risks there will be no delegation."

"It is one of the favorite pastimes of headquarters groups to decide from within their professional ivory tower what help the field organization needs and to design and develop programs for meeting these "needs." Then it becomes necessary to get field management to accept the help provided, and a different role is taken by the staff: that of persuading middle and lower management to utilize the programs."

"Man will exercise self-direction and self-control in the service of objectives to which he is committed."

"The approach that dominates organizational theory, teaching, and practice for most of the twentieth century looked at organizations from the top-down, starting with a view of the CEO as the "leader" who shapes the organization's strategy, structure, culture, and performance potential. The nature of work and the role of the workforce enter the analysis much later, after considerations of technology and organization design have been considered. However, if the key source of value in the twenty-first-century organization is to be derived from the workforce itself, an inversion of the dominant approach will be needed. The new perspective will start not at the top of the organization, but at but at the front lines, with people and the work itself — which is where value is created. Such an inversion will lead to a transformation in the management and organization of work workers, and knowledge. This transformation was signalled by McGregor, but we must go further."

"[Douglas McGregor] coined the two terms and used them to label two sets of beliefs a manager might hold about the origins of human behaviour. He pointed out that the manager's own behaviour would be largely determined by the particular beliefs that he subscribed to....McGregor hoped that his book would lead managers to investigate. "But that isn't what happened. Instead McGregor was interpreted as advocating Theory Y as a new and superior ethic - a set of moral values that ought to replace the values managers usually accept."

"The approach that dominates organizational theory, teaching, and practice for most of the twentieth century looked at organizations from the top-down, starting with a view of the CEO as the "leader" who shapes the organization's strategy, structure, culture, and performance potential. The nature of work and the role of the workforce enter the analysis much later, after considerations of technology and organization design have been considered. However, if the key source of value in the twenty-first-century organization is to be derived from the workforce itself, an inversion of the dominant approach will be needed. The new perspective will start not at the top of the organization, but at the front lines, with people and the work itself — which is where value is created. Such an inversion will lead to a transformation in the management and organization of work workers, and knowledge. This transformation was signalled by McGregor, but we must go further."

"This paper develops a new theoretical model with which to examine the interaction between technology and organizations. Early research studies assumed technology to be an objective, external force that would have deterministic impacts on organizational properties such as structure. Later researchers focused on the human aspect of technology, seeing it as the outcome of strategic choice and social action. This paper suggests that either view is incomplete, and proposes a reconceptualization of technology that takes both perspectives into account. A theoretical model-- the structurational model of technology--is built on the basis of this new conceptualization, and its workings explored through discussion of a field study of information technology. The paper suggests that the reformulation of the technology concept and the structurational model of technology allow a deeper and more dialectical understanding of the interaction between technology and organizations. This understanding provides insight into the limits and opportunities of human choice, technology development and use, and organizational design. Implications for future research of the new concept of technology and structurational model of technology are discussed."

"Technology has always been a central variable in organizational theory, informing research and practice. Despite years of investigative effort there is little agreement on the definition and measurement of technology, and no compelling evidence on the precise role of technology in organizational affairs. I will argue that the divergent definitions and opposing perspectives associated with technological research have limited our understanding of how technology interacts with organizations, and that these incompatibilities cannot be resolved by mutual concession. What is needed is a reconstruction of the concept of technology, which fundamentally re-examines our current notions of technology and its role in organizations."

"Two views on the scope of technology have pervaded (and shaped) studies of technology, reflecting the different claims to generalizability that researchers have intended with their work. The one set of studies has focused on technology as "hardware," that is, the equipment, machines, and instruments that humans use in productive activities, whether industrial or informational devices."

"Rather than positing design and use as disconnected moments or stages in a technology's lifecycle, the structurational model of technology posits artifacts as potentially modifiable throughout their existence. In attempting to understand technology as continually socially and physically constructed, it is useful to discriminate analytically between human action which affects technology and that which is affected by technology. I suggest that we recognize human interaction with technology as having two iterative modes: the design mode and the use mode. I emphasize that this distinction is an analytical convenience only, and that in reality these modes of interaction are tightly coupled."

"Technology is built and used within certain social and historical circumstances and its form and functioning will bear the imprint of those conditions."

"As both technologies and organizations undergo dramatic changes in form and function, organizational researchers are increasingly turning to concepts of innovation, emergence, and improvisation to help explain the new ways of organizing and using technology evident in practice. With a similar intent, I propose an extension to the structurational perspective on technology that develops a practice lens to examine how people, as they interact with a technology in their ongoing practices, enact structures which shape their emergent and situated use of that technology. Viewing the use of technology as a process of enactment enables a deeper understanding of the constitutive role of social practices in the ongoing use and change of technologies in the workplace. After developing this lens, I offer an example of its use in research, and then suggest some implications for the study of technology in organizations."

"Technology - and its relationship to organizational structures, processes, and outcomes - has long been of interest to organizational researchers. Over the years, different research perspectives on technology have developed in parallel with research perspectives on organizations - for example, contingency theory (Woodward 1965, Galbraith 1977, Carter 1984, Daft and Lengel 1986), strategic choice models (Child 1972, Buchanan and Boddy 1983, Davis and Taylor 1986, Zuboff 1988), Marxist studies (Braverman 1974, Edwards 1979, Shaiken 1985, Perrolle 1986), symbolic interactionist approaches (Kling 1991, Prasad 1993), transaction-cost economics (Malone et al. 1987, Ciborra 1993); network analyses (Barley 1990, Burkhardt and Brass 1990, Rice and Aydin 1991), practice theories (Suchman 1987, Button 1993, Hutchins 1995, Orr 1996), and structurational models (Barley 1986, Orlikowski 1992, DeSanctis and Poole 1994)."

"Cellular automata are now being used to model varied physical phenomena normally modelled by wave equations, fluid dynamics, Ising models, etc. We hypothesize that there will be found a single cellular automaton rule that models all of microscopic physics; and models it exactly. We call this field DM, for digital mechanics."

"Under the roof of one controversial assumption about physics, we discuss five big questions that can be addressed using concepts from a modern understanding of digital informational processes. The assumption is called finite nature. The digital mechanics model is obtained by applying the assumption to physics. The questions are as follows: 1. What is the origin of spin? 2. Why are there symmetries and CPT (charge conjugation, parity, and time reversal)? 3. What is the origin of length? 4. What does a process model of motion tell us? 5. Can the finite nature assumption account for the efficacy of quantum mechanics?"

"Feynman considered Fredkin a brilliant and consistently original, though sometimes incautious, thinker. If anyone is going to come up with a new and fruitful way of looking at physics, Feynman said, Fredkin will."

"Bennett from IBM, Fredkin, and later Toffoli investigated whether, with gates that are reversible, you can do everything. And it turns out, wonderfully true, that the irreversibiilty is not essential for computation. It just happens to be the way we designed the circuits."

"The black hole information paradox is probably the most important issue for fundamental physics today. If we cannot understand its resolution, then we cannot understand how quantum theory and gravity work together. Yet very few people seem to understand how robust the original Hawking arguments are and what exactly it would take to resolve the problem."

"The exact description of any physical process must include the effects of quantum gravity. But our experience suggests that there is a separation of scales, so that under suitable conditions we get 'lab physics'; i.e., physics described to good accuracy by quantum fields on gently curved spacetime."

"In the fuzzball paradigm, the black hole microstates have no interior, and radiate unitarily from their surface through quanta of energy E ∼ T. But quanta with E ≫ T impinging on the fuzzball create large collective excitations of the fuzzball surface. The dynamics of such excitations must be studied as an evolution in superspace, the space of all fuzzball solution |Fi⟩."

"This (machine learning algorithm that can predict SARS-CoV-2 virus mutations) is a real-time companion to vaccine development. What we can do with our model right now is a lot faster than what you can do in the lab."

"Basically the American modernization (of nuclear weapons), and Russia’s unfortunate inability to improve their early-warning system, has resulted in a situation where everything is potentially a lot more dangerous, because an accident could much more easily occur. And this is both a social, political and technical problem."

"[Putin's] afraid of the misinformed American president doing something that gets everybody killed. He’s not worried about us getting ourselves killed, but he is worried about Russia. So what he wants to do is make it clear to anybody — to a child on a bicycle — that you cannot win. They will destroy the United States in response, no matter what your defenses can or cannot do."

"We’re talking about a wall of fire that encompasses everything around us at the temperature of the center of the sun. That will literally turn us to less than ash, if this thing gets going. I can’t emphasize how powerful these weapons are. When they detonate, they’re actually four or five times hotter than the center of the sun, which is 20 million degrees Kelvin. They’re 100 million degrees Kelvin at the center of these weapons."

"A fetus is no more a human than an acorn is a tree"

"Many women of my generation, in many fields, had good reason to be grateful to the women's colleges."

"In the wake of Pearl Harbor, a single word favored above all others by Americans as best characterizing the Japanese people was "treacherous," and for the duration of the war the surprise attack on the U.S. Pacific fleet remained the preeminent symbol of the enemy's inherent treachery. The attack also inspired a thirst for revenge among Americans that the Japanese, with their own racial blinders, had failed to anticipate. In one of his earliest presentations of the plan to attack Pearl Harbor, even Admiral Yamamoto Isoroku, who presumedly knew the American temperament firsthand from his years as a naval attache in Washington, expressed hope that shattering opening blow against the Pacific Fleet would render both the U.S. Navy and the American people "so dispirited that they will not be able to recover.""

"Even after they were checked at Midway and Guadalcanal in 1942, many Japanese remained convinced that the Anglo-American enemy was indeed psychologically incapable of recovering In actuality, the contrary was true, for the surprise attack provoked a rage bordering on the genocidal among Americans. Thus, Admiral William Halsey, soon to become commander of the South Pacific Force, vowed after Pearl Harbor that by the end of the war the Japanese language would be spoken only in hell, and rallied his men thereafter under such slogans as "Kill Japs, kill Japs, kill more Jap." Or as the U.S. Marines put it in a well-known variation on Halsey's motto: "Remember Pearl Harbor- keep 'em dying.""

"Indeed, in wartime jargon, the notion of "good Japanese" came to take on an entirely different meaning than that of "good Germans," as Admiral William F. Halsey emphasized at a news conference early in 1944. "The only good Jap is a Jap who's been dead for six months," the commander of the U.S. South Pacific Force declared, and he did not mean just combatants. "When we get to Tokyo, where we're bound to get eventually," Halsey went on, "we'll have a little celebration where Tokyo was." Halsey was improvising on a popular wartime saying, "the only good Jap is a dead Jap," and his colleagues in the military often endorsed this sentiment in their own fashion."

"Among the Allied war leaders, Admiral Halsey was the most notorious for making outrageous and virulently racist remarks about the Japanese enemy in public. Many of his slogans and pronouncements bordered on advocacy of genocide. Although he came under criticism for his intemperate remarks, and was even accused of being drunk in public, Halsey was immensely popular among his men and naturally attracted good press coverage. His favorite phrase for the Japanese was "yellow bastards," and in general he found the color allusion irresistible."

"Black Lives Matter was a piece of genus called declarative marketing. I don't know if you've ever heard of it via products in the 1970's called "Gee, Your Hair Smells Terrific" or "I Can't Believe It's Not Butter!" was the name of the product. So the name of the product is called Black Lives Matter. How can you disagree with that?"

"Who came up with ethno-nationalism? [Is] swedes caring about Sweden ethno-nationalism? FU! You know, people have a right to be in their country without somebody saying "Oh, that's just blood and soil like from the nazis.""

"There is a jewish strategy. The great part of the jewish strategy is that most of it is pretty much open source. If you want to push your children really, really hard to survive and if you want to tell them "You've got a dragon breathing fire down the back of your neck because you've always been oppressed and you never know when you have to leave very quickly on short notice", you can duplicate the jewish experience. Good luck!"

"If Ashoka founders think it was a company because they put so much money in it, let us now talk business. Many in the leading US universities were thinking of developing partnerships with Ashoka and a lead founder came to me for a partnership with Brown. Not possible anymore."

"The academy never stood apart from American slavery—in fact, it stood beside church and state as the third pillar of a civilization built on bondage."

"In the early days of this project, software was treated like an adopted child and not taken as seriously as other engineering disciplines, such as hardware engineering, and was thought of as art and magic, not science. I have always believed that art and science were involved in its creation, but at the time most people thought otherwise. Knowing this, I fought to legitimize software so that both software engineering and those who built it would receive the respect they deserved, so I began using the term “software engineering” to differentiate it from hardware and other forms of engineering. When I first started using these words, they were considered funny. It was a running joke for a long time. They liked to make fun of my radical ideas. Software eventually earned the same respect as any other discipline."

"...I still remember those girls and who didn't love math, uh, the way I did. And I mean, oh my gosh, think what they were missing. I know I always felt, you know, solving math problems was a little bit like eating candy. There was something about it. It was so rewarding. It was just such a pleasure to do it. And I thought, oh, once they see this, they're going to enjoy it too."

"...And it was so ridiculous. And this was this thing about this teaching of a class and I've been told that I couldn't teach undergraduates because MIT students didn't believe scientific information spoken by a woman. And so I'd said, well, of course, everyone knows that. I had accepted it as normal because as soon as somebody said it, I realized, of course, it's true. I was able to see that women were so under-respected that students couldn't respect them enough. And so they were afraid to put an important course into the hands of a woman for fear the students would not be able to respect them."

"The notion that global warming is a fact and will be catastrophic is drilled into people to the point where it seems surprising that anyone would question it, and yet, underlying it is very little evidence at all. Nonetheless, there are statements made of such overt unrealism that I feel embarrassed. I feel it discredits science. I think problems will arise when one will need to depend on scientific judgment, and by ruining our credibility now you leave society with a resource of some importance diminished."

"With respect to science, the assumption behind consensus is that science is a source of authority and that authority increases with the number of scientists. Of course, science is not primarily a source of authority. Rather, it is a particularly effective approach to inquiry and analysis. Skepticism is essential to science; consensus is foreign."

"We're talking of a few tenths of a degree change in temperature. None of it in the last eight years, by the way. And if we had warming, it should be accomplished by less storminess. But because the temperature itself is so unspectacular, we have developed all sorts of fear of prospect scenarios – of flooding, of plague, of increased storminess when the physics says we should see less. I think it's mainly just like little kids locking themselves in dark closets to see how much they can scare each other and themselves."

"Based on the weak argument that the current models used by the IPCC couldn't reproduce the warming from about 1978 to 1998 without some forcing, and that the only forcing that they could think of was man. Even this argument assumes that these models adequately deal with natural internal variability—that is, such naturally occurring cycles as El Niño, the Pacific Decadal Oscillation, the Atlantic Multidecadal Oscillation, etc. Yet articles from major modeling centers acknowledged that the failure of these models to anticipate the absence of warming for the past dozen years was due to the failure of these models to account for this natural internal variability. Thus even the basis for the weak IPCC argument for anthropogenic climate change was shown to be false."

"If we just take the default view today, people will be going to these places with an extractive mindset that says- I have the technology, money and power, and I’ll use these resources until I’m satisfied, and I will not be concerned with other countries and future generations."

"If we don't take a radical shift toward really prioritizing labor rights, it's quite concerning imagining having a company or a government controlling the full life support system on a space station or a physical base on a planet."

"Space truly is useful for sustainable development for the benefit of all peoples."

"Many of today’s successful applications of artificial intelligence (AI) in medical imaging focus on the automation of tasks that radiologists can do."

"As the field matures, I hypothesise that AI models will be able to answer many questions that are challenging for physicians."

"What is the risk of getting a future disease."

"What is the efficacy of a certain treatment."

"How is a certain disease going to progress."

"I cannot imagine healthcare without a human physician in the loop."

"While AI imaging tools can certainly bring many benefits, they are not perfect. Their best utilisation will depend on how well they are integrated into the clinical pipeline."

"It is up to human experts to design safe and effective protocols."

"I predict that as these imaging models continue to develop, a lot of low-level tasks will be delegated to AI."

"The ultimate responsibility for clinical decisions will remain with physicians."

"To bring AI-imaging tools into a clinical setting, it is essential to have the agreement of physicians."

"AI is not part of the curriculum in most medical schools."

"This lack of background makes it challenging for physicians to adopt and trust this new technology."

"I think that in cancer and in many other diseases."

"The big question is always, how do you deal with uncertainty."

"It's all the matter of predictions."

"Today, we rely on humans who don't have this capacity to make predictions."

"Many times people get wrong treatments or they are diagnosed much later."

"Algorithms govern our computing-based world in the same way that the laws of nature govern the physical one."

"Their mathematical underpinnings are thus as important to modern society."

"As the periodic table, relativity or the genome."

"The Simons Institute at Berkeley, under my leadership, will continue its dedication to the discovery of the fundamentals of computation and to findings that enable technological progress and positive social change."

"I am very proud to have won the Turing Award."

"Our work was very unconventional at the time."

"We were graduate students and let our imagination run free, from using randomized methods to encrypt single bits to enlarging the classical definition of a proof."

"To allow a small error to setting new goals for security."

"Winning the award is further testimony to the fact that the cryptographic and complexity theoretic community embraced these ideas in the last 30 years."

"AI and physics is really a two-way street: AI’s influence on how we research new physics phenomena, but also applying physics thinking to the way that AI systems operate."

"I basically told them, in no uncertain terms, that this was not the way that I thought physics research should be going. And they agreed with me, actually, about the off-the-shelf use of machine learning."

"Where would someone live within the academic ecosystem if they wanted to dive into physics topics, but also computation and statistics? That was the motivation for submitting a grant to the NSF to start this institute.If you think about AI solely as it applies to fundamental physics research, we have massive challenges in our field and we’re trying to understand some of the deepest questions in nature."

"Cosmology is another area with massive amounts of data and massive computational costs to run simulations of the universe. Without something like machine learning, we simply wouldn't be able to tackle those problems."

"there is physics for AI, where there are concepts from physics that can help you build better AI tools, even if you aren’t directly studying physical systems. You can build machine learning architectures that have physics-style reasoning embedded and those concepts turn out to be quite powerful."

"Invoking the simple principle of translational symmetry — which in nature gives rise to conservation of momentum — led to dramatic improvements in image recognition"

"Computational biology is the art of developing and applying computational methods to answer questions in biology, such as studying how proteins fold, identifying genes that are associated with diseases, or inferring human population histories from genetic data."

"I have interests in both the development of computational methods and in answering specific biology questions, primarily related to the function of RNA, a molecule central to the function of cells."

"While genome sequencing has obviously been useful in revealing the sequences that are involved in coding various aspects of the molecular biology of the cell, it has had a secondary impact that is less obvious at first glance."

"The low cost and high throughput (the ability to process large volumes of material) of genome sequencing allowed for a more "big-data" approach to biology, so that experiments that previously could only be applied to individual genes could suddenly be applied in parallel to all of the genes in the genome."

"A result of the scale of these new experiments is the emergence of very large data sets in biology whose interpretation demands the application of state-of-the-art computer science methods."

"The problems require interdisciplinary dexterity and involve not only management of large data sets but also the development of novel abstract frameworks for understanding their structure."

"OK, so what's my research goal? I come from the machines end of this world, roughly. And what I really want to do is figure out how it is that we can make intelligent robots. And I do this mostly because I'm interested in intelligence more than I'm interested actually in robots. But I think that trying to make a physical agent who goes out and interacts in the world is a really good test bed for understanding what kinds of reasoning and perception and control we need in order to make it an intelligent system."

"So the way I think about the problems-- this is kind of a definitely a computer scientist way to think about the problem-- is to think about the robot as a transducer, as some kind of a system that's connected up to the world. And it makes observations of the world. And it takes actions that change the state of the world. And presumably, there's some objective, right? We want to take actions that change the state of the world in some way that we think will be good."

"The reason I want to start by backing all the way up to this like very basic control theory picture is that right now there's an enormous amount of argument about how one should make robots. Should they do planning and reasoning? Should they do reinforcement learning? How should we do it? So there's a huge kind of crisis almost in the field about what the best methods are. And what I want to start out this talk by doing is actually thinking about how we can answer that question in a way that's not political or religious, but technical."

"So the way I want to think about this, the job of this program. So I'm going to make a robot. I'm going to put a program in the head of the robot. So let's say, I'm not going to worry about hardware. I'm just going to read about the software. And so the program that I'm going to put in the head of my robot, it has to do this job that's written in the formula up here. And what this is just shorthand for saying is that it has to represent some kind of mapping from observation and actions that it's had in the past. So o, a star means the whole history of observations and actions that it's ever had. Based on that, it has to pick the next action. So that's not really saying much of anything at all. That's just a description of every single robot control program basically that's been written. You have to take your history of actions and observations, compute the next action."

"And so what we want to do is think about first of all, what's the best-- what would be the best pi to put inside the robot. How can we think about that? And then we have to think about the problem of how is it that we, in my case, as me, as an engineer, I'm going to find that pi that I should put in my robot."

"So one way to think about the whole problem set up then is that I, as the robotics engineer, have to do for my robots the job that nature did for you. That is to say, I have to think about I'm a robot factory. I'm going to make these robots. And the robots are going to go out in the world. Maybe they're going to go and work in people's Kitchens or something. And every kitchen is going to be different. So there's going to be a lot that I don't know about the world. But somehow, I have to figure out the best program, what program, to put in the head of all my robots, so that when they go out in the world to behave, they can do a good job so that's the way I think about the problem that I face. And in order to think about what would be the best program, I kind of think about it this way."

"So I imagine that there's some distribution over possible environments that the robot could find itself in when it actually goes out into the world, right? So maybe it's going to go to houses and the houses are all somewhat different. And once I put that program in the house, maybe it's going to do some estimation or learning. It's going to adapt to the circumstances it's in. My job is to find a program that does a good job of adapting in all the environments that might find itself in."

"So imagine that you have some kind of probability distribution over the worlds that the robot could actually end up operating in. I want to find a program that's going to behave well, let's say get a lot of reward in expectation on average over all the environments that it could possibly find itself in. So that's, I would say, kind of a reasonable formal objective for a robot. And one thing that's good about this as an objective is that we don't have to argue about it, right? It doesn't say whether there should be learning in there or what kind of learning or should it be a genetic algorithm or should it have planning. In some sense, you could say, "I just want to make the program that's going to be the best that can be on average over these environments.""

"But the problem is now I've written down an objective function. I've said, "Oh, if you could tell me a distribution over possible worlds that you'd like this program to work well in, then I know in a certain mathematical sense with the best program is." But now my problem as the engineer, as the person who is in the robot factory, which is again, the kind of maybe analogous to the problem of nature, is I have to figure out how do I how do I find this program that's going to be good and all these situations?"

"So there are a bunch of ways you can think about the problem. I mean, one would be to say, "Oh, I'm really lazy. I don't really want to think very much about working in the factory. It seems awfully hard. I will just make a robot that has roughly an empty head. It doesn't really know very much at all. And then it just has to interact in the world and learn everything by interacting." But of course, you don't really want a robot that comes to your kitchen and begins to learn about physics, right? That would break a lot of dishes."

"Another strategy-- and this is like the classic engineering strategy-- is that, no, I'm like a serious engineer. And I'm going to sit here and think really, really hard. And I'm going to write a program. And it's going to be a great program. And I'm just going to put it straight in the robot's head. And it's going to go off, and it's going to be awesome and do everything it needs to do. And that strategy actually can work very well in certain kinds of problems. It lets that, the Boston dynamics robots do Parkour. But as we try to address bigger and more complicated problems, it becomes harder and harder for engineers to just straight up write the program."

"We could just try to figure out how humans work because humans work pretty well in a variety of domains. And so one program would be to say, "Well, we forget how humans work. And then that's what we do. We make robots that work like that." So first of all, that's a hard biology problem. I think it's very important that people work on it. But it's also not a general engineering methodology because for instance, I might want robots that work in certain kinds of circumstances or problem domains that are really different from the niche that humans are well tuned for. And so I might want to make a robot that isn't really human-like in its intelligence. And then it seems like what we're left with that maybe we could just say, well, we'll somehow recapitulate evolution. Like we just search around in the space of programs and try to find ones that work well and then eventually get ones that are great for our environment. But that seems slow and complicated."

"So if I enumerate my options and they all don't look very good, I don't know what to do. So one thing to think about, though, is this last thing. So the kind of evolution idea. So let's just pursue this a little bit more. So imagine that we want to try to find a program that works well in expectation over all environments. One way to think about that is that inside the factory, we kind of simulate a bunch of environments. We try a bunch of robot programs. And we try to find one that works well in all those environments. And that's like a really interesting strategy. We would have to think of a space of possible programs for the robot, some objective function. We figure out, well, what are we trying to optimize, a distribution over problems to test."

"In some sense, this is a thing that people have thought about for a long time, right? This would be like running some kind of evolutionary algorithm or some search or simulation inside the factory. And it's very attractive, but I think generally speaking, hard to make work well. So the question is what should I do, right? I could maybe I can set up this whole evolutionary setup somehow. And then I could just snooze for a really long time while some very complicated program tries to figure out the best robot program to put in the head of the robot. But I don't know. I am simultaneously too impatient for that."

"And so then the question is can I somehow take pieces and parts of all these ideas, some human programming, some robot learning in the wild, some kind of search or evolution offline, some inspiration from humans. Can I take all those things and put them together and see if I can find a way to engineer intelligent robots? So that's basically what I'm up to."

"I'm going to-- well, no. OK, let me say something about this. So then the one way to view the research agenda is to say that first of all, I'd like to be inspired by what we know about humans. And in particular, I'm very interested in this bulky core knowledge type stuff because that tells me something about what evolution, in some sense, saw fit to engineer into natural intelligences. And if I understand that natural systems seem to be born with a bias or some built in structure to think in terms of other agents, to understand that they move through 3D space, to talk about, think about objects as clumps of matter that cohere, that's a very helpful engineering bias for building a system."

"I also know just some physics and variance about the worlds that my robot's going to operate in. And maybe humans don't have this built in explicitly, but they almost surely have a built in implicitly. And I also have some other constraints as an engineer who's trying to make intelligent robots, which is that humans are the engineers, right?"

"So if humans have to engineer a very complicated system, then it has to be the engineering process has to have some modularity to it because humans are really bad at understanding one big messy system. They're good at understanding pieces and parts that work together. So it may be that we have to take a modular design approach in our engineering efforts for intelligence, not because the intelligence needs to have that architecture, but because we, the human engineers, need those tools for actually building a system."

"So all these constraints need to somehow come together into a way of building intelligence systems. Actually, I would stop here for a minute just because it's a convenient spot and see if there are questions. I see some red Q&A button. So maybe someone can ask."

"Yeah, actually. For years there has been. So a more typical formalization would be in terms of predictive models and planning or reasoning. So reinforcement learning. Also, it depends. The phrase unfortunately, the phrase, "reinforcement learning," grows and stretches too. And sometimes for many people and in many discourses, it's come to mean all of intelligent behavior, in which case, I would say, well, no it's all reinforcement learning. But that's vacuous. Other formulations involve reasoning about objects and their relationships and thinking about the long term consequences of taking actions in the world and so on. So there's really different ways of framing and formalizing the problem. And they give you very different computational profiles and different learning strategies. OK, good."

"So I'll just tell you some story because people usually like stories, and it's kind of the afternoon. So and this is related to the question about reinforcement learning, probably, right? So how did I get into this whole thing? When I just finished my undergraduate degree, which actually was in philosophy, weirdly enough, I went to work at a research institute while I was starting my PhD. And they had this robot nobody really knew actually very much about robotics. So And it was my job as the brand new person to try to get the robot to drive down the hallway."

"And so what happened was I programmed the robot. And it would run into the wall. And I would bring it back. And I would fix the programming. And it would run into the wall again, hopefully for a slightly different reason. And over the course of a couple of weeks, I managed to write a program that would use these funny sonar sensors on the robot and make it drive down the hall without crashing into the walls. And so that was good. And I was happy, in a way, at the end of that, that I had gotten it to work. But I reflected on that a bit more. And what I decided was that I had learned how to navigate down the hallway using the sonar sensors. And what I thought was that-- and it had taken a long time. And I was kind of a hassle. And really, the system should have been doing the learning, not me. So my view was that I should figure out a way to get out of the loop to build systems that could learn on their own to do stuff. And then I could just wait for them to do that. And that would be better. So that was flaky."

"Then I sort of reinvented reinforcement learning in a not very good way, really. But it was kind of entertaining. And I this is a slide by the way for those young people in the audience. You might know, but back in the day, we used to write with colored depends on pieces of clear plastic. And that's what we used to give talks. So I had this kind of pseudo reinforcement learning thing. And by 1990, I actually had this little robot called Spanky that did actual reinforcement learning during my actual defense. So it didn't learn anything too complicated. But it did do it in real time. So that was kind of fun."

"So OK, I finished my PhD. And I thought, OK, I know something about robot learning now. But I really want to make robots that can do complicated things. And I couldn't figure out how to get basic reinforcement learning methods to really scale up to problems that I cared about. And so this is one last flight. I'll show you from some talk that I gave in 1995. And I kind of complained that the ideal that you could take just a big bunch of what I like to call neural goo now, just a big bunch of generic neural network stuff, and train it to be an intelligent agent all by itself. But that wasn't going to be feasible. And instead, we needed some kind of compositional structure. And that would give us more efficient learning and more robust behavior and so on."

"So I'm still there, OK? So I'm still in I'm still trying to figure out how we can design an architecture that can learn efficiently. And so the research strategy that I have really adopted, I work closely with a colleague, Tomás Lozano-Pérez. Our strategy has been the following, which is to try to think of some very generic representation and inference mechanisms and build those in and then figure out how to learn the rest of the stuff. And we're all used to I think by now the idea of some representation that inference mechanisms that we would want to build in."

"For instance, everyone's used to the idea of convolution now in image space. But if you think about it. And I've had people tell me who work on convolutional neural networks that they don't build any structure into their system. It's just a neural network. But of course, as soon as you build the compositional structure into a neural network, you are taking a position on some regularities that are in the input signal and so on. And you're taking advantage of that so that you don't have to learn a whole fully connected network, but you just learn some compositional kernels."

"So just as convolution gives us a great leverage when you apply it to the right part of the problem, then the intuition as well, hopefully there there's a few mechanisms, hopefully not like 100 mechanisms, but maybe 10. And then if we figure out how to use those mechanisms to bias learning and to structure behavior that we could learn robust ways of behaving that are efficient and so on."

"So one set of possible kind of general ideas includes convolution in space also and time. Maybe understanding the kinematics of the system that it's connected together in joints and segments, a notion of planning to move through space, being able to do causal reasoning-- if I were to do this, what would happen? -- Abstracting over individual objects, various kinds of state and temporal abstraction and so on. So our view-- I don't want to commit to a particular list-- but is that there's a list of structural principles that are pretty generic and very broadly useful and we should build them in."

"OK, I'll keep going. I'll surely be able to offend some people soon. And I'll work harder at that. OK. so if we kind of accept this idea that we're going to build in some structure, then what? And the thing that my colleague and I have done recently. Well, now, maybe not super recently, but recent. In order to test out the idea that there's a set of mechanisms that would work well, what we did is we hand built the rest of the system. So we hand-built some transition models, inference rules, ways of doing search control and so on and connected them up to these general mechanisms and made a system."

"And just again, to kind of give you the motivation. I really want a robot. This isn't my kitchen, by the way, just in case you were worried, my kitchen. But imagine that you had to clean this kitchen or make breakfast in it or something. It would be very hard. And imagine programming a robot to do it. It's extremely hard. And so one thing that's useful to do is to think about what makes this problem hard. So one of the things that makes it hard is that there are lots of objects."

"So the dimensionality of the space is kind of unthinkably high. It's also not exactly clear what constitutes an object here. If you were going to behave in this world, it would be a very long sequence of primitive actions that you would take in order to clean this kitchen. And also there's just a fundamental amount of uncertainty in this problem, right? So you don't know what's in the blue bowl or what will happen if you try to pull out a certain thing. You don't know when the people are coming home or what they want for dinner all sorts of stuff you don't know."

"And so any approach that works effectively in a domain like this is going to have to handle very large spaces, very long horizons, and really lots of uncertainty. So we have kind of a standard structural decomposition to this problem. We call this belief space hierarchical planning in the now. I'll decode what that means a little bit."

"Fundamentally, the way we think about it is that we decompose the computation that's in the robot's head now into two parts. The first part is in charge of taking the sequence, the history of actions and observations, and trying to synthesize them into some representation of a belief for a probability distribution about the way the world might be and then another module that takes that belief and decides how to behave."

"I didn’t allow myself to work on it during the day,” she said, “because I didn’t consider it to be real math. I thought it was, like, my homework."

"It’s really [three- and four-dimensional shapes] that are exciting for me, but the study of these things is deeply linked with knot theory."

"We can have our econ ‘hats’ on with some theory in mind, but without a deep understanding of the context and constraints in the particular environment you're working in, it's hard to design interventions that might be useful"

"Everyone has their own path to finding their passion and, for some, that may be longer than for others."

"Yeah, absolutely. I’m actually Kenyan, by the way. So I was born and raised there, which is why I do research there. But yes, I’ve been to them several times. In fact, I was just there, I was there in February. Yeah. I mean, they’re not rich people, right? This is rural Kenya. They’re pretty poor. They’re mostly growing and eating maize. That’s the main staple of the whole country. A lot of them might not eat much meat a year, maybe at Christmas."

"Every step in my career has served a purpose. Of course, each step provided me with a particular level of training. But the steps also bought me time to figure out what I like and what I don’t like professionally and personally. If I had to end with some advice, I would say: Take each of these steps seriously and not to rush ahead to the next one."

"When you don't feel valued, you feel like sh*t. When you feel like sh*t, you don't do good math."

"I grew up not having any clue what a Mathematician was… so I looked up and wrote to random Math Professors. It sounds silly, but this actually helped shape my career path. Doing math for a living is a sweet gig, and I am deeply grateful to be a part of the mathematics community."

"Over the past several years, we’ve seen Driver Monitoring Systems (DMS) go from nice to have to must have technology in cars."

"In cabin sensing is the natural evolution of DMS and is seeing increased demand because of its many important applications in safety as well as entertainment, comfort, wellness and beyond. Instead of just focusing on the driver, interior sensing solutions have a full view of the entire cabin. With a camera and other sensors, these systems provide human-centric insight into what’s happening in a vehicle, by detecting the state of the cabin and that of the driver and passengers in it."

"In cabin sensing is such a hot topic because it allows OEMs to not only meet regulatory and rating requirements that are on the horizon, but also sets them up to differentiate their brands in a very competitive market. Also, car manufacturers can expand on cameras, other sensors and machine learning-based algorithms that are already deployed for DMS. Building on that as a foundation, they can often, cost effectively, add advanced safety features and engaging mobility experiences to existing platforms."

"Our focus should always be on advancing automotive technologies to save lives and I love how interior sensing enables several advanced safety features. One important functionality is child and child seat detection, which helps determine if a child is left behind in a vehicle unattended. I am hopeful that this can help avoid tragic deaths due to vehicular heatstroke. And, as anyone who has driven around with kids or for that matter, pets will know, they often cause distraction, so child and pet detection can provide important inputs to promote safer driving behavior."

"I have always been very interested in the intersection of healthcare and automotive. So, I am glad to see that, for example through Euro NCAP, attention is being given to sudden sickness. What if interior sensing AI could detect driver impairment due to a medical event? For example, we already have early stage capabilities for detecting heart rate variability using computer vision, and I am eager to see these types of technologies advance so that they can actually be deployed in vehicles."

"Occupancy detection, which determines how many people are in the cabin and where they are sitting, is another interesting area of functionality. This too can help improve existing safety functionality, as it helps determine proper seat position and seat belt usage and can provide important analysis for airbag deployment."

"I am also very interested in the experience features that in cabin sensing will enable. The Emotion AI technology that we created at Affectiva, now part of Smart Eye, also unlocks massively interesting personalization. By understanding the emotional and cognitive states of people in a vehicle we can adapt the environment to their needs in the moment. For example, if someone is getting drowsy, the lighting dims, the heating turns up and soothing music plays quietly in the background, creating a comfortable and restorative environment."

"There are also fascinating backseat use cases around content recommendations and monetization of advertising data, even. Understanding passenger reactions and emotional engagement with music and video content can help refine content recommendations, making these more relevant to the user and their experience. Where advertising is deployed, understanding viewer engagement with that content provides OEMs and advertisers with very valuable data. I think users will be interested in opting in to that, if there is an incentive for them. Imagine a scenario where you and your friends take a rideshare to a concert and engage with advertising in the back of a car. In return, since the system knows you are taking a ride to a concert, it offers you a discount coupon for a band t-shirt or such. And, to wrap this up, a fun use case often requested by OEMS, is trip highlights where the Emotion AI detects smiles and joy, so that an interior camera can take cool pictures just at the right moment there’s a lot of fun we can add to the overall experience considering the right applications of this technology."

"Integrity is essential. It's the inner voice, the source of self-control, the basis for the trust that is imperative in today's military. It's doing the right thing when nobody's looking."

"If women don't belong in engineering, then engineering, as a profession, is irrelevant to the needs of our society. If engineering doesn't make welcome space for them, then engineering will become marginalized as other fields expand their turf to seek out and make a place for women."

"It's not easy being a pioneer. It's not easy having to prove every day that you belong. It's not easy being invisible or having your ideas credited to someone else."

"Aircraft trailing vortices have little waves that are generated and then break up....If another aircraft intercepts that trailing vortex, someone can be killed, because it’s a swirling flow....People now know that if you land airplanes three minutes apart, that’s going to be safer."

"Friends may come and go, but enemies accumulate."

"The spectacular results in the fluctuation theory of sums of independent random variables, obtained in the last 15 years by , , , , , , and others, have gradually led to the realization that the nature of the problem, as well as that of the methods of solution, is algebraic and combinatorial. After Baxter showed that the crux of the problem lay in simplifying a certain operator identity, several algebraic proofs (, , Wendel) followed. It is the present purpose to carry this algebraization to the limit: the result we present amounts to a solution of the for s. The solution is not presented as an algorithm, but by showing that every identity in a Baxter algebra is effectively equivalent to an identity of symmetric functions independent of the number of variables. Remarkably, the identities used so far in the combinatorics of fluctuation theory "translate" by the present method into classical identities of easy verification. The present method is nevertheless also useful for guessing and proving new combinatorial identities: by way of example, it will be shown in the second part of this note how it leads to a generalization of the Bohnenblust-Spitzer formula for the action of arbitrary ."

"It has been observed that whereas s and s are likely to be embarrassed by references to the beauty in their work, mathematicians instead like to engage in discussions of the beauty of mathematics. Professional artists are more likely to stress the technical rather than the aesthetic aspects of their work. Mathematicians, instead, are fond of passing judgment on the beauty of their favored pieces of mathematics. Even a cursory observation shows that the characteristics of mathematical beauty are at variance with those of artistic beauty. For example, courses in “art appreciation” are fairly common; it is however unthinkable to find any “mathematical beauty appreciation” courses taught anywhere. The purpose of the present paper is to try to uncover the sense of the term “beauty” as it is currently used by mathematicians."

"It cannot be a complete coincidence that several outstanding logicians of the twentieth century found shelter in s at some time in their lives: Cantor, , Gödel, Peano, and are some. was one of the saner among them, though in some ways his behavior must be classified as strange, even by mathematicians' standards. He looked like a cross between a and a large owl. He spoke softly in complete paragraphs which seemed to have been read out of a book, evenly and slowly enunciated, as by a . When interrupted, he would pause for an uncomfortably long period to recover the thread of the argument. He never made casual remarks: they did not belong in the baggage of ."

"The more experimental scientists and s are, the more common sense they have, and so on until you get to the mathematicians, who are totally devoid of common sense."