Eye On A.I.

In the second episode of a series on the machine-learning pipeline, Joe Hellerstein, a professor at UC Berkeley, talks about data wrangling.

Turing Award winner David Patterson talks about the end of Moore's law and the evolution of computer chips from general purpose CPUs to GPUs to the recent Cambrian explosion of specialized AI chips, including Google's TPU, in the first episode of five about the machine-learning model-development pipeline with Ameet Talwalkar, co-founder of Determined AI, an open-source platform for ML model development.

NSCAI commissioner Gilman Louie talks about great power competition in the digital age and the US need for a more technologically literate diplomatic core to counter China's increasing diplomatic efforts on the technology front.

Ken Ford, an NSCAI commissioner, talks about turning AI from a tool to a teammate and unshackling research and implementation of AI from government bureaucracy in order to compete with nimbler countries such as China.

Katharina McFarland, an NSCAI commissioner, talks about the commission's recommendations on inserting AI into the National Defense Strategy and how war games should integrate AI enabled applications in order to inform US military doctrine

Eric Horvitz, Chief Scientific Officer at Microsoft and a commissioner with the National Security Commission on AI, talks about the commission's recommendations that responsible and ethical uses be a primary consideration in any AI system for national security.

Jason Matheny, an NSCAI commissioner who now leads a think tank at Georgetown University focused on the intersections of technology, policy and national security, spoke in this week's episode about the Commission's recommendations on export controls for sensitive technology and the difficult problem of protecting digital assets like algorithms and data.

Mignon Clyburn, an NSCAI commissioner and longtime regulator at the FCC, talks about the Commission's recommendations to increase competency in artificial intelligence at the Department of Defense and other government agencies. Ms. Clyburn talked about establishing a National Reserve Digital Corps with a scholarship program modeled after the ROTC and the creation of a United States Digital Service Academy similar to mint a new class of civil servants with deep technical knowledge.

The Alexa Prize Socialbot Grand Challenge is a competition for university students to create a social bot that can converse coherently and engagingly with humans. This year's prize goes to the team from Emory University in Atlanta, Georgia. I speak with Prem Natarajan, vice president of natural understanding in the Alexa AI organization, about the prize, the evolution of conversational AI, its current challenges and its future promise

Qualcomm Technologies is powering the shift to fifth generation wireless communications, known as 5G. In this sponsored episode, I speak to two of the company's product managers leading the company's efforts to use AI to enhance 5G, reduce on-device power consumption, and give developers the tools necessary to make devices smarter and faster.

This week I speak to Lieutenant General Jack Shanahan, recently retired Director of the Pentagon's Joint Artificial Intelligence Center,or JAIC. He was instrumental in starting Project Maven to integrate state-of-the-art computer vision into drone technology. He then started the JAIC, the central hub for the military's AI efforts. Gen. Shanahan spoke about the challenges of nurturing innovation within a rigid and multilayered organization like the DOD and the threats the US faces ahead.

This week I speak to Stanford professor Fei-Fei Li, one of the people responsible for the current AI revolution. Fei-Fei talked about her early days running a New Jersey dry cleaner to finance her Princeton education, her creation of ImageNet, the world's first large labeled image data set, which allowed the validation of neural networks, and her latest work on ambient intelligence, which promises to transform elder care.

Pietro Perona, a professor at the California Institute of Technology, is one of the brains behind a pair of smartphone apps that help you do just that: eBird and iNaturalist. But simple as the apps are to use in identifying tens of thousands of species, the science behind them is complex. Gathering enough examples for each species to train a neural network is impossible, so much of Prof. Perona's work has been focused on making machines more efficient learners, requiring less training data.

In this week's episode, the last in a series of six, we talked to Eric Horvitz. Commissioner at the National Security Commission on AI and Chief Research Scientist at Microsoft, about the commission's recommendations to Congress about the need for training, standards and documentation to govern the application of AI in the national security space.

In this week's episode, we talk with NSCAI Commissioner Gilman Louie about line of effort's recommendations to Congress on how to maintain US leadership in AI hardware and 5G. He spoke about the need for a national microelectronic strategy and policies that would advance spectrum sharing for 5G to create a global alternative to the Chinese telecommunications giant Huawei.

NSCAI Commissioner Jason Matheny talks about his line of effort regarding cooperation on AI with key allies and partners. He spoke about his group's recommendations that the government establish a senior national security point of contact for AI and convene a multilateral working group for AI collaboration and interoperability - as well as the need for AI war games.

In the third episode of six episodes looking at the National Security Commission on AI 's first quarter recommendations to Congress, we speak with Katharina McFarland, a former Assistant Secretary of Defense for Acquisition, about the commission's recommendation that the Department of Defense and the Office of the Director of National Intelligence establish a steering committee on emerging technology to ensure that AI for national security gets top priority in the years ahead.

In this week's episode, we speak to Jose-Marie Griffiths, a commissioner on the National Security Commission on Artificial Intelligence, about the commission's recommendations on how to strengthen the federal government's AI workforce. The recommendations focused on raising understanding of AI within the government and the need to streamline government hiring practices in order to attract and retain talent.

In April 2020, the National Security Commission on Artificial Intelligence issued its first-quarter recommendations to Congress, covering seven lines of effort, six of which are public and one of which is classified. In the first of a series of podcast episodes about those recommendations, we spoke with Andrew Moore, a professor at Carnegie Mellon University and NSCAI commissioner about the commission's recommendations on increased AI R&D funding.

COVID-19 continues to sweep through the human population, killing some and damaging the health of others. While this podcast is normally focused on machine-learning, this week I talk to Vittorio Sebastiano, an assistant professor of stem cell biology at Stanford University, about groundbreaking tech that could someday help restore scarred tissue to pre-COVID health. Vittorio talked about his hunt for machine-learning collaborators to understand the process further.

COVID-19 has swept across the world was startling speed, but with equally startling speed, the machine learning community has responded. This week I speak with Irina Rish, a professor at the University of Montreal and a Mila academic member, who is helping head a task force to understand the virus. She talked about where the efforts currently stand and where they expect to go in the weeks and months ahead. Let me know when it's live.

There has been a debate in the past few years between the symbolists and the connectionists about the future of artificial intelligence. The symbolists say that traditional, explainable, logic-based approaches still hold tremendous promise while the connectionists say that the power of deep learning, for all its current opacity and narrow application, holds the key to more general forms of machine intelligence. This week, I speak with David Cox, IBM Director of the MIT-IBM Watson AI Lab, which is blending the two traditions in what they call neuro-symbolic AI in hopes to move AI forward.

Justin Gottschlich, who founded the machine programming research group at Intel Labs, explains his group's efforts to automate software development. The ambition is to make it possible for anybody to create software simply by describing what they intend the software to do.

This week I talk to Casimir Wierzynski, a senior director in Intel's AI Products Group, Cas talked about his work in privacy, taking me on a tour of the latest strategies that promise to unlock the data necessary to liberate AI. He talked about hardening encryption against the code-cracking power of quantum computers and about his work in connectomics with salami slicers for the brain that are making it possible to map the neural networks of our minds.

Terry Sejnowski, author of the book Deep Learning Revolution, who together with Geoff Hinton created Boltzmann machines, a deep learning network that has remarkable similarities to learning in the brain, talks about whether machines dream and the algorithms of the brain, whether Marvin Minsky was the devil and how deep learning is shaping the future of education.

We begin 2020 by looking back at some of the highlights from 2019 including conversations with Turing award winners, Yoshua, Bengio and Yann Lecun, as well as with the father of reinforcement learning, Rich Sutton. Our guests consider applying machine learning to the climate crisis; competition between the U S and China for dominance in AI; and the future of machine learning through various kinds of unsupervised learning.

Daphne Koller, formerly at Stanford University and cofounder of the online education company, Coursera, talks this week about using machine-learning to develop new drugs. Her approach is to use machine learning to accuratley identify cellular or genetic targets for treatment. The field is just getting started but promises to speed the development of new and better therapies to treat disease.

My guest this week, Aude Billard from Switzerland's Learning Algorithms and Systems Laboratory, blends control theory with machine learning to build robotic systems that are both swift and precise but can handle some of the unpredictability of the real world. Her lab famously taught a robot arm to catch a tennis racket looping through the air and is working on ever more precise robots that can even do the work of Switzerland's famous watchmakers.

Former Google chief executive Eric Schmidt and former Deputy Defense Secretary Bob Work, co-chairs of the U.S. National Security Commission on AI, talk about the challenges the government faces in winning support from a skeptical private sector and in maintaining engagement with China while ensuring that that engagement doesn't work to America's detriment.

The secret in much of artificial intelligence today is that it depends on hordes of unskilled workers to label the data used to train supervised learning models. But, in order for data science teams to work with labelers around the world, they need a platform. This week, in the second of a periodic series of sponsored episodes, I talk to Manu Sharma and Brian Rieger, who saw the opportunity to provide that platform and founded Labelbox, the leading labelling software in the space.

This week, I talk to Dawn Song, one of the world's foremost experts in computer security, about her vision of a new paradigm in which people control their data and are compensated for its use by corporations. Dawn, a professor at the University of California, Berkeley, has recently launched a company, Oasis Labs, which is building a platform that brings together the immutability of blockchain and the privacy of secure enclaves to give data owners the ability to control their data.

A few months ago at the recent international conference on machine learning, a workshop and research paper launched a movement to use machine learning in addressing climate change. The response was huge and has given birth to the bones of an organization climate change.ai. This week I talked to David Rolnick, a postdoc at U Penn and Priya, Donti, a Phd student at Carnegie Mellon, about how the group came together and about how the organization is developing.

Automated machine-learning tools – or tools that automate the creation of machine-learning applications – are increasingly important in the current talent-scarce environment. Expensive ML engineers shouldn't spend their time doing stuff that machines can do quicker and cheaper. This week, we talk to Evan Sparks and Ameet Talwalkar, two of the founders of Determined AI, which builds tools that streamline workflows for machine-learning teams and, the company hopes, will eventually democratize AI.

This week, I talk to Brendan McCord, who wrote the Pentagon's AI strategy and is now a Special Government Employee at the National Security Commission on AI. Brendan talks about what he believes the US needs to do to stay competitive with China and promote an alternative vision of AI-powered security and prosperity to the world.

This week I talk to John Platt, a Distinguished Scientist at Google, about twin problems: finding cheap zero-carbon energy sources and mitigating global warming. John is a polymath, having discovered asteroids, helped put the touch in computer touchpads and even won an Academy Award for scientific and technical achievements in computer animation. Now, he is part of a growing movement of machine learning researchers tackling climate change.

This week we return to the world of thinking robots with Chelsea Finn, one of the youngest experts in the field, who talks about her journey, about her work in meta-learning and about lifelong learning for robots.

This week, we look at AI in India. With its massive population, fast-growing economy, English-language education and large supply of brilliant researchers and engineers, it should be competing with China and the U.S. for dominance in the space. But it is not. I talk to Partha Talukdar, a professor at the Indian Institute of Science, Bangalore, about the challenges that have kept India from realizing its AI potential.

This week I talk to Yann Lecun, one of the brightest minds in machine learning today. Yann's work lies behind some of the most critical AI applications, most notably computer vision systems that power everything from face recognition software to self-driving cars. He recently won the Turing Award, the highest prize in computer science. We talked about Yann's first computer, about how music led him into computer science, and about his work on self-supervised learning, which he believes will take us to human-level intelligence in machines.

This week I talk to Trae Stephens and Brian Schimpf from Anduril Industries, an AI defense contractor, about the current state of AI research and deployment for national security, including how the US stacks up against China. We also talked about the resistance among US engineers to work on defense applications and whether that hobbles the US in the global AI arms race.

This week, I talk to Ken Church, a pioneer in Natural Language Processing, whose use of statistical models on part of speech tagging revolutionized the field and is what makes automatic dictation and machine translation so popular today. We talked about his early days at MIT, about explainable AI and about how the Holy See played a role in his probabilistic approach to NLP.

This week, I talk to Sergey Levine, one of the most prolific researchers in robot learning. We talked about developing a robot's sense of touch and about robot dreams and whether he believes we know what's happening in the field in Russia and China.

Thinking robots: that's how much of the world envisions artificial intelligence and if there is one person on the planet who understands the limitations and promise of intelligence in robots, it's Pieter Abbeel, one of the world's foremost experts on robotic learning systems. In this episode, Pieter talks about robot memories and the prospect of robots with personalities eventually assisting in the home. Listen and learn about your future.

This week I talk to the Bengio brothers, Samy and Yoshua, in their first interview together. Yoshua recently won the Turing Award with Geoff Hinton and Yann Lecun, while Samy leads a team of researchers at Google Brain. The brothers are well known to people who work in machine learning, but few know how intertwined their professional lives have been. They talked about their unconventional parents and their early collaboration on neural network research, as well as what they see as the challenges ahead.

In this week's episode, I talk to Pedro Domingos, author of the bestselling book, The Master Algorithm, which is about the ongoing effort to unify machine-learning paradigms in a single model. But the conversation was much broader than that. Pedro believes strongly that the great powers are engaged in an AI arms race with America's Defense Advanced Research Projects Agency, or Darpa, pitted against China's military and industrial dynamo. We also talked about the future of democracy and authoritarianism in an AI-driven world.

Resurrecting the Tower of Babel with machine learning: In this week's episode, we talk to Liang Huang, principal scientist at Baidu Research in Silicon Valley about his breakthrough in simultaneous translation technology, which promises to erase language barriers. Baidu's system can already translate speech to text with as little as a three-second delay. Soon, Liang says, the technology will translate speech to speech, enabling a future in which people from different languages can speak fluidly with each other.

In this episode of Eye on AI, I continue my review of AI research in different regions of the world with a focus on Europe. Europe, with its strong academic tradition, faces unique challenges in scientific research because of the continent's fragmentation and the weakening of the European Union. Given the growing dominance of North America and China, Europe risks being left behind. To understand what Europe is doing to avoid this, I talk to Bernhard Schölkopf and Matthias Bethge, two machine learning researchers from Tübingen, Germany, who have been working to bring machine learning research in Europe under one umbrella. I hope you find Bernhard and Matthias as interesting as I did.

In this episode of Eye on AI, I talk to Ben Rosman, who runs Africa's largest machine learning lab at the University of Witwatersrand in Johannesburg, South Africa, where he is focused on reinforcement learning. Ben has been instrumental in unifying and upgrading the continent's machine-learning capabilities in hopes of making Africa a player in the artificial intelligence revolution. If you wanted to know what's going on with AI in Africa, Ben will fill you in. I hope you find him as interesting as I did.

In this episode of Eye on AI, I talk to Julian Togelius, perhaps the most prolific researcher at the intersection of video games and artificial intelligence. Julian works on AI for games and games for AI. Some of his most significant work is in training deep neural networks to play video games and generalize what they have learned, a critical step toward artificial general intelligence. For those of you who don't understand the importance of video games to artificial intelligence research, Julian will enlighten you. For those of you who do his recent work will surprise you. I hope you find Julian as interesting as I did.

In this episode of Eye on AI, I talk to Miles Brundage, who studies the societal impacts of artificial intelligence and works on the policy team of OpenAI, the nonprofit A.I. research company founded by Elon Musk. When I spoke to Miles, he was a research fellow at the University of Oxford's Future of Humanity Institute, where he remains an associate. We talked about the policy side of AI security and whether he is optimistic that regulations can steer machine learning applications away from the nightmare scenarios popularly imagined. I hope you find Miles as interesting as I did.