from the world's big
The biggest problem in AI? Machines have no common sense.
Correlation doesn't equal causation — we all know this. Well, except robots.
Dr. Gary Marcus is the director of the NYU Infant Language Learning Center, and a professor of psychology at New York University. He is the author of "The Birth of the Mind," "The Algebraic Mind: Integrating Connectionism and Cognitive Science," and "Kluge: The Haphazard Construction of the Human Mind." Marcus's research on developmental cognitive neuroscience has been published in over forty articles in leading journals, and in 1996 he won the Robert L. Fantz award for new investigators in cognitive development.
Marcus contributed an idea to Big Think's "Dangerous Ideas" blog, suggesting that we should develop Google-like chips to implant in our brains and enhance our memory.
GARY MARCUS: The dominant vision in the field right now is, collect a lot of data, run a lot of statistics, and intelligence will emerge. And I think that's wrong. I think that having a lot of data is important, and collecting a lot of statistics is important. But I think what we also need is deep understanding, not just so-called "deep learning." So deep learning finds what's typically correlated, but we all know that correlation is not the same thing as causation. And even though we all know that and everybody learned that in Intro to Psych, or should have learned that in Intro to Psych they should have learned that you don't know whether cigarettes cause smoking just from the statistics we have. We have to make causal inferences and do careful studies. We all know that causation and correlation are not the same thing.
Have right now as AIs-- giant correlation machines. And it works, if you have enough control of the data relative to the problem that you're studying that you can exhaust the problem, to beat the problem into submission. So you can do that with go. You could play this game, over and over again, the rules never change. They haven't changed in 2,000 years. And the board is always the same size. And so you can just get enough statistics about what tends to work, and you're good to go. But if you want to use the same techniques for natural language understanding, for example, or to guide a domestic robot through your house, it's not going to work. So the domestic robot in your house is going to keep seeing new situations. And your natural language understanding system, every dialogue is going to be different. It's not really going to work. So yeah, you can talk to Alexa and you can say, please turn on my over and over again, get statistics on that. It's fine. But there's no machine in the world T=that we're having right now. It's just not anywhere near a reality and you're not going to be able to do it with the statistics, because there's not enough similar stuff going on.
Probably the single best thing that we could do to make our machines smarter is to give them common sense, which is much harder than it sounds like. I mean, first you might say, what is common sense? And what we settled on in the book that we wrote is that common sense is the knowledge that's commonly held. that ordinary people have. And yet machines don't. So machines are really good at things like, I don't know, converting metrics -- you know, converting from the English system to the metric system. Things that are nice, and precise, and factual, and easily stated. But things that are a little bit less sharply stated, like how you open a door, machines don't understand the first thing. So there's actually a competition right now for opening doors. So if somebody uploaded data sets from 500 different doors, and they're hoping that robots will experiment with all 500 doors and then get it. But what's more likely is that they'll get to door 501 and they'll actually have a problem or at least door 601. So every ordinary person in the west has opened a bunch of door knobs and gets the idea, right? I need to turn something, jiggle something -- might be different for the next one -- until the door itself is free.
So you can give a definition of something like that. It's hard to give a perfect definition. But we all know that. And yet nobody's ever built a robot that can do that. We made a joke about it in the book. We said, you know, in the event of a robot attack do the following six things and number one was close the door. And we added that you might need to lock it. And that was before this whole database came out. So, you know, the field advances. People are working on that. Maybe next year they'll work on teaching robots about locks. But I bet you it will take a while before robots understand all the little ways it can jiggle the key, and maybe you need to pull in the door to make it just right. We don't know how to even encode that information in a language that a computer can understand. So the big challenge of common sense is to take stuff like that-- like, how do you open the door, why do you want to open a door-- and translate into the language of the machine. It's a lot harder than a metric converter. It's a lot harder than a database. And right now the field's not even really trying to answer that question. It's so obsessed with what it can do with these big databases, which are exciting in themselves, that it's kind of lost sight of that, even though the question itself that goes back to the 1950s, when one of the founders of A.I., John McCarthy, first started writing about it in the late '50s.
But it's not-- common sense is not getting the attention that it deserves and that's one of the reasons we wrote the book. Common sense is just one step along the way to intelligence. People talk about artificial intelligence. And sometimes they talk about artificial general intelligence. There's also narrow AI. And narrow AI is the stuff that we're doing pretty well now. So, do the same problem over and over again, just solving one problem. You could think about idiot Savants that can do a calendar but can't do anything else and can tell you what day you were born on if you give them your birthday. We have a lot of narrow AI right now. We can't do narrow AI for everything that we want to do. But the dream is to have broad AI, or general AI, that can solve any problem. You think about the Star Trek computer. You can say, you know, please give me the demographics in this galaxy and cross-correlate it with this, and with that, and tell me this. And Star Trek computer says, OK. And it figures it out. So the Star Trek computer understands everything about language, and it understands pretty much everything about how the world works. And it can put those together to give you an answer.
It's not like Google, right? Google can search for pages that have the information. But then you have to put the information together. The Star Trek computer can synthesize it. And it's a great example of general intelligence. If I ask you, did George Washington have a cell phone? Well, if you have a kind of common knowledge of when cellphones were introduced, when George Washington was alive, the fact that he's dead, then you could compute the information and give me the right answer. If you Google for it, you might get a wacky answer. If it works for George Washington, maybe you search for, did Thomas Jefferson have a cell phone? And the answer should obviously be the same. But maybe that one won't be on a Google page. So if you have common sense about things like time, and space, and a lot of sort of everyday factual knowledge, that gets you a long way to intelligence. It doesn't give you all the way there. So general intelligence, first of all, has many dimensions to it. So you can think like, the SAT has verbal and math.
Those are two of the of intelligence that's Really not well-established right now in the AI community is common sense. There are other aspects of intelligence, like doing pure calculation, where the A.I. community has done a good job. But there are other things that go into intelligence as well. For example, the ability to read a graph is partly about common sense, and it's partly about understanding what people might intend, it's partly sometimes about expert knowledge. So reading a graph is another form of intelligence. And that's going to require putting together better perceptual tools than we have now, better common sense tools, probably some knowledge about politics, for example, if you're reading a graph that's relevant to the latest political campaign, and so forth. So there's a lot of stuff Right now we're trying to approximate it all with statistics. But it's never general knowledge. So you could learn to read one graph, Going help you read the next graph. So general intelligence is going to be putting together a lot of things, both some that we already understand pretty well in the A.I. community and some we just haven't been working on and really need to get back to.
- There are a lot of people in the tech world who think that if we collect as much data possible, and run a lot of statistics, that we will be able to develop robots where artificial "intelligence" organically emerges.
- However, many A.I.'s that currently exist aren't close to being "intelligent," it's difficult to even program common sense into them. The reason for this is because correlation doesn't always equal causation — robots that operate on correlation alone may have skewed algorithms in which to operate in the real world.
- When it comes to performing simple tasks, such as opening a door, we currently don't know how to encode that information — the varied process that is sometimes required in differing situations, i.e. jiggling the key, turning the key just right — into a language that a computer can understand.
Gary Marcus is the author of Rebooting AI: Building Artificial Intelligence We Can Trust.
- Human-like AI could emerge in 5 to 10 years - Big Think ›
- The next Stage of Machine Learning: Teach Robots to Think Like ... ›
- What animals is A.I. currently smarter than? - Big Think ›
Higher education faces challenges that are unlike any other industry. What path will ASU, and universities like ASU, take in a post-COVID world?
- Everywhere you turn, the idea that coronavirus has brought on a "new normal" is present and true. But for higher education, COVID-19 exposes a long list of pernicious old problems more than it presents new problems.
- It was widely known, yet ignored, that digital instruction must be embraced. When combined with traditional, in-person teaching, it can enhance student learning outcomes at scale.
- COVID-19 has forced institutions to understand that far too many higher education outcomes are determined by a student's family income, and in the context of COVID-19 this means that lower-income students, first-generation students and students of color will be disproportionately afflicted.
What conditions of the new normal were already appreciated widely?<p>First, we understand that higher education is unique among industries. Some industries are governed by markets. Others are run by governments. Most operate under the influence of both markets and governments. And then there's higher education. Higher education as an "industry" involves public, private, and for-profit universities operating at small, medium, large, and now massive scales. Some higher education industry actors are intense specialists; others are adept generalists. Some are fantastically wealthy; others are tragically poor. Some are embedded in large cities; others are carefully situated near farms and frontiers.</p> <p>These differences demonstrate just some of the complexities that shape higher education. Still, we understand that change in the industry is underway, and we must be active in directing it. Yet because of higher education's unique (and sometimes vexing) operational and structural conditions, many of the lessons from change management and the science of industrial transformation are only applicable in limited or highly modified ways. For evidence of this, one can look at various perspectives, including those that we have offered, on such topics as <a href="https://www.insidehighered.com/digital-learning/blogs/rethinking-higher-education/lessons-disruption" target="_blank">disruption</a>, <a href="https://www.nytimes.com/2020/02/20/education/learning/education-technology.html" target="_blank">technology management</a>, and so-called "<a href="https://www.insidehighered.com/sites/default/server_files/media/Excerpt_IHESpecialReport_Growing-Role-of-Mergers-in-Higher-Ed.pdf" target="_blank">mergers and acquisitions</a>" in higher education. In each of these spaces, the "market forces" and "market rules" for higher education are different than they are in business, or even in government. This has always been the case and it is made more obvious by COVID-19.</p> <p>Second, with so much excitement about innovation in higher education, we sometimes lose sight of the fact that students are—and should remain—the core cause for innovation. Higher education's capacity to absorb new ideas is strong. But the ideas that endure are those designed to benefit students, and therefore society. This is important to remember because not all innovations are designed with students in mind. The recent history of innovation in higher education includes several cautionary tales of what can happen when institutional interests—or worse, <a href="https://www.insidehighered.com/news/2016/02/09/apollos-new-owners-seek-fresh-start-beleaguered-company" target="_blank">shareholder</a> interests—are placed above student well-being.</p>
Photo: Getty Images<p>Third, it is abundantly apparent that universities must leverage technology to increase educational quality and access. The rapid shift to delivering an education that complies with social distancing guidelines speaks volumes about the adaptability of higher education institutions, but this transition has also posed unique difficulties for colleges and universities that had been slow to adopt digital education. The last decade has shown that online education, implemented effectively, can meet or even surpass the quality of in-person <a href="https://link-springer-com.ezproxy1.lib.asu.edu/article/10.1007/s10639-019-10027-z" target="_blank">instruction</a>.</p><p>Digital instruction, broadly defined, leverages online capabilities and integrates adaptive learning methodologies, predictive analytics, and innovations in instructional design to enable increased student engagement, personalized learning experiences, and improved learning outcomes. The ability of these technologies to transcend geographic barriers and to shrink the marginal cost of educating additional students makes them essential for delivering education at scale.</p><p>As a bonus, and it is no small thing given that they are the core cause for innovation, students embrace and enjoy digital instruction. It is their preference to learn in a format that leverages technology. This should not be a surprise; it is now how we live in all facets of life.</p><p>Still, we have only barely begun to conceive of the impact digital education will have. For example, emerging virtual and augmented reality technologies that facilitate interactive, hands-on learning will transform the way that learners acquire and apply new knowledge. Technology-enabled learning cannot replace the traditional college experience or ensure the survival of any specific college, but it can enhance student learning outcomes at scale. This has always been the case, and it is made more obvious by COVID-19.</p>
What conditions of the new normal were emerging suspicions?<p>Our collective thinking about the role of institutional or university-to-university collaboration and networking has benefitted from a new clarity in light of COVID-19. We now recognize more than ever that colleges and universities must work together to ensure that the American higher education system is resilient and sufficiently robust to meet the needs of students and their families.</p> <p>In recent weeks, various commentators have suggested that higher education will face a wave of institutional <a href="https://www.businessinsider.com/scott-galloway-predicts-colleges-will-close-due-to-pandemic-2020-5" target="_blank">closures</a> and consolidations and that large institutions with significant online instruction capacity will become dominant.</p> <p>While ASU is the largest public university in the United States by enrollment and among the most well-equipped in online education, we strongly oppose "let them fail" mindsets. The strength of American higher education relies on its institutional diversity, and on the ability of colleges and universities to meet the needs of their local communities and educate local students. The needs of learners are highly individualized, demanding a wide range of options to accommodate the aspirations and learning styles of every kind of student. Education will become less relevant and meaningful to students, and less responsive to local needs, if institutions of higher learning are allowed to fail. </p> <p>Preventing this outcome demands that colleges and universities work together to establish greater capacity for remote, distributed education. This will help institutions with fewer resources adapt to our new normal and continue to fulfill their mission of serving students, their families, and their communities. Many had suspected that collaboration and networking were preferable over letting vulnerable colleges fail. COVID-19's new normal seems to be confirming this.</p>
President Barack Obama delivers the commencement address during the Arizona State University graduation ceremony at Sun Devil Stadium May 13, 2009 in Tempe, Arizona. Over 65,000 people attended the graduation.
Photo by Joshua Lott/Getty Images<p>A second condition of the new normal that many had suspected to be true in recent years is the limited role that any one university or type of university can play as an exemplar to universities more broadly. For decades, the evolution of higher education has been shaped by the widespread imitation of a small number of elite universities. Most public research universities could benefit from replicating Berkeley or Michigan. Most small private colleges did well by replicating Williams or Swarthmore. And all universities paid close attention to Harvard, Princeton, MIT, Stanford, and Yale. It is not an exaggeration to say that the logic of replication has guided the evolution of higher education for centuries, both in the US and abroad.</p><p>Only recently have we been able to move beyond replication to new strategies of change, and COVID-19 has confirmed the legitimacy of doing so. For example, cases such as <a href="https://www.washingtonpost.com/education/2020/03/10/harvard-moves-classes-online-advises-students-stay-home-after-spring-break-response-covid-19/" target="_blank">Harvard's</a> eviction of students over the course of less than one week or <a href="https://www.nhregister.com/news/coronavirus/article/Mayor-New-Haven-asks-for-coronavirus-help-Yale-15162606.php" target="_blank">Yale's apparent reluctance</a> to work with the city of New Haven, highlight that even higher education's legacy gold standards have limits and weaknesses. We are hopeful that the new normal will include a more active and earnest recognition that we need many types of universities. We think the new normal invites us to rethink the very nature of "gold standards" for higher education.</p>
A graduate student protests MIT's rejection of some evacuation exemption requests.
Photo: Maddie Meyer/Getty Images<p>Finally, and perhaps most importantly, we had started to suspect and now understand that America's colleges and universities are among the many institutions of democracy and civil society that are, by their very design, incapable of being sufficiently responsive to the full spectrum of modern challenges and opportunities they face. Far too many higher education outcomes are determined by a student's family income, and in the context of COVID-19 this means that lower-income students, first-generation students and students of color will be disproportionately afflicted. And without new designs, we can expect postsecondary success for these same students to be as elusive in the new normal, as it was in the <a href="http://pellinstitute.org/indicators/reports_2019.shtml" target="_blank">old normal</a>. This is not just because some universities fail to sufficiently recognize and engage the promise of diversity, this is because few universities have been designed from the outset to effectively serve the unique needs of lower-income students, first-generation students and students of color.</p>
Where can the new normal take us?<p>As colleges and universities face the difficult realities of adapting to COVID-19, they also face an opportunity to rethink their operations and designs in order to respond to social needs with greater agility, adopt technology that enables education to be delivered at scale, and collaborate with each other in order to maintain the dynamism and resilience of the American higher education system.</p> <p>COVID-19 raises questions about the relevance, the quality, and the accessibility of higher education—and these are the same challenges higher education has been grappling with for years. </p> <p>ASU has been able to rapidly adapt to the present circumstances because we have spent nearly two decades not just anticipating but <em>driving</em> innovation in higher education. We have adopted a <a href="https://www.asu.edu/about/charter-mission-and-values" target="_blank">charter</a> that formalizes our definition of success in terms of "who we include and how they succeed" rather than "<a href="https://www.washingtonpost.com/opinions/2019/10/17/forget-varsity-blues-madness-lets-talk-about-students-who-cant-afford-college/" target="_blank">who we exclude</a>." We adopted an entrepreneurial <a href="https://president.asu.edu/read/higher-logic" target="_blank">operating model</a> that moves at the speed of technological and social change. We have launched initiatives such as <a href="https://www.instride.com/how-it-works/" target="_blank">InStride</a>, a platform for delivering continuing education to learners already in the workforce. We developed our own robust technological capabilities in ASU <a href="https://edplus.asu.edu/" target="_blank">EdPlus</a>, a hub for research and development in digital learning that, even before the current crisis, allowed us to serve more than 45,000 fully online students. We have also created partnerships with other forward-thinking institutions in order to mutually strengthen our capabilities for educational accessibility and quality; this includes our role in co-founding the <a href="https://theuia.org/" target="_blank">University Innovation Alliance</a>, a consortium of 11 public research universities that share data and resources to serve students at scale. </p> <p>For ASU, and universities like ASU, the "new normal" of a post-COVID world looks surprisingly like the world we already knew was necessary. Our record breaking summer 2020 <a href="https://asunow.asu.edu/20200519-sun-devil-life-summer-enrollment-sets-asu-record" target="_blank">enrollment</a> speaks to this. What COVID demonstrates is that we were already headed in the right direction and necessitates that we continue forward with new intensity and, we hope, with more partners. In fact, rather than "new normal" we might just say, it's "go time." </p>
Hollywood has created an idea of aliens that doesn't match the science.
- Ask someone what they think aliens look like and you'll probably get a description heavily informed by films and pop culture. The existence of life beyond our planet has yet to be confirmed, but there are clues as to the biology of extraterrestrials in science.
- "Don't give them claws," says biologist E.O. Wilson. "Claws are for carnivores and you've got to be an omnivore to be an E.T. There just isn't enough energy available in the next trophic level down to maintain big populations and stable populations that can evolve civilization."
- In this compilation, Wilson, theoretical physicist Michio Kaku, Bill Nye, and evolutionary biologist Jonathan B. Losos explain why aliens don't look like us and why Hollywood depictions are mostly inaccurate.
Sallie Krawcheck and Bob Kulhan will be talking money, jobs, and how the pandemic will disproportionally affect women's finances.
Manly Bands wanted to improve on mens' wedding bands. Mission accomplished.
- Manly Bands was founded in 2016 to provide better options and customer service in men's wedding bands.
- Unique materials include antler, dinosaur bones, meteorite, tungsten, and whiskey barrels.
- The company donates a portion of profits to charity every month.
Scientists uncovered the secrets of what drove some of the world's last remaining woolly mammoths to extinction.
Every summer, children on the Alaskan island of St Paul cool down in Lake Hill, a crater lake in an extinct volcano – unaware of the mysteries that lie beneath.