AI with AI: Artificial Intelligence with Andy Ilachinski

Andy and Dave discuss the latest in AI news, including a report that the Israel Defense Forces used a swarm of small drones in mid-May in Gaza to locate, identify, and attack Hamas militants, using Thor, a 9-kilgram quadrotor drone. A paper in the Journal of American Medical Association examines an early warning system for sepsis, and finds that it misses out on most instances (67%) of cases, and frequently issued false alarms (to which the developer contests the results). A new bill, the Consumer Safety Technology Act, directs the US Consumer Product Safety Commission to run a pilot program to use AI to help in safety inspections. A survey from FICO on The State of Responsible AI (2021) shows, among other things, a disinterest in the ethical and responsible use of AI among business leaders (with 65% of companies saying that can’t explain how specific AI model predictions are made, and 22% of companies have an AI ethics board to consider questions on AI ethics and fairness). In a similar vein, a survey from the Pew Research Center and Elon University’s Imagining the Internet Center found that 68% of respondents (from across 602 leaders in the AI field) believe that AI ethical principles will NOT be employed by most AI systems within the next decade; the survey includes a summary of the respondents’ worries and hopes, as well as some additional commentary. GitHub partners with OpenAI to launch CoPilot, a “Programming Partner” that uses contextual cues to suggest new code. Researchers from Stanford University, UC San Diego, and MIT research Physion, a visual and physical prediction benchmark to measure predictions about commonplace real world physical events (such as when objects: collide, drop, roll, domino, etc). CSET releases a report on Machine Learning and Cybersecurity: Hype and Reality, finding that it is unlikely that machine learning will fundamentally transform cyber defense. Bengio, Lecun, and Hinton join together to pen a white paper on the role of deep learning in AI, not surprisingly eschewing the need for symbolic systems. Aston Zhang and Zack C. Lipton, and Alex J Smola release the latest version of Dive into Deep Learning, now over 1000 pages, and living only as an online version.

Follow the link below to visit our website and explore the links mentioned in the episode.

Andy and Dave discuss the latest in AI news and research, including, the UK government releases its National AI Strategy, a 10-year plan to make the country a global AI superpower [1:28]. Stanford University’s One Hundred Year Study on AI Project releases its second report, Gathering Strength, Gathering Storms, assessing developments in AI between 2016 and 2021 around fourteen framing questions. [4:57] The UN High Commissioner for Human Rights calls for a moratorium on the sale and use of AI systems that pose series risks to human rights until adequate safeguards are put into place. [10:07] Jack Poulson at Tech Inquiry maps out US government use of AI-based weapons and surveillance, using publicly available information. [12:07] Researchers at Hebrew University examine the potential of single cortical neurons as deep artificial neural networks, finding that a deep neural network with 5-8 layers are necessary to approximate them. [16:10] Researchers at Stanford review the different architectures of neuronal circuits in the human brain, identifying different circuit motifs. [20:02] Other research at Stanford shows the ability to image and track moving non-line-of-sight objects using a single optical path (shining a laser through a keyhole). [22:05] And researchers at MIT, Nvidia, and Technion demonstrate that a neural network can identify the number and activity of people in a room, solely by examining a blank wall in the room. [26:33] The Nils Theory research group publishes Physics-Based Deep Learning, introducing physical models into deep learning to reconcile data-centered viewpoints with physical simulations. [30:34] Ori Cohen compiles the Machine and Deep Learning Compendium, an open resource (GitBook) on over 500 topics with summaries, links, and articles. [32:21] The Allen Institute for AI releases a web tool that converts PDF papers into HTML for more rapid web publishing of scientific papers. [33:20] And the Museum of Wild and Newfangled Art: This Show is Curated by a Machine invites viewers to ponder on why they think an AI chose the works within. [34:43]

In news items, Andy and Dave discuss an effort by Boston Children’s Hospital to use machine learning to help track the spread of COVID-19. Meanwhile, a proposal from researchers wants to use mobile phones to track the virus’s spread. Fifty-two organization have come together to develop the “first-ever industry-led” standard for AI in healthcare. The National Oceanic and Atmospheric Administration (NOAA) announces its AI strategy. And IBM and Promare begin sea trials for Mayflower, an autonomous ship that, later this year, will make the reverse of the 1620 Mayflower transit, completely unmanned. In research, Google and Columbia University enable a robot to teach itself how to walk with minimal human intervention (bounding the terrain, and making the robot’s trial movements more cautious). Researchers at Harvard, MIT CSAIL, IBM-Watson-AI Lab, and DeepMind introduce CLEVRER (Collision Events for Video Representation and Reasoning), a diagnostic video dataset for the evaluation of models on a wide range of reasoning tasks. And DeepMind proposal a new reinforcement learning technique that models human behavior, using a gifting game in which agents learn to trust each other. The Berkman Klein Center at Harvard updates its data map of Ethical and Rights-based approaches to Principles for AI. The Center for the Study of the Dragon releases its likely last paper, Unarmed and Dangerous, which looks at how non-weaponized drones can still have lethal effects. Cansu Canca has provided a database and interface that looks at global dynamics of AI principles. Mario Alemi provides the book of the week, with the Amazing Journey of Reason: from DNA to AI. And the livestream talks from the 34th AAAI Conference are now available online.

Click here to visit our website and explore the links mentioned in the episode. 

This week, Microsoft Research and University of Montreal show that machines can learn through interactive language by answering questions (question answering with interactive text, or QAit). The Allen Institute for AI’s Aristo system, a suite of eight solves, can pass (90%+) the New York 8th Grade regents science exams (for non-diagram, multiple choice questions), and can exceed 83% on the 12th grade exam, though Melanie Mitchell suggests the achievement may not be as profound as it seems. A “meta-research” paper from Milan and Klagenfurt takes a broader look at neural network research and highlights concerns of reproducibility (or lack thereof) as well as utility (or lack thereof, where simple heuristic methods can outperform the neural networks). From a workshop organized by Max Tegmark and Emilia Javorsky, a group of diverse authors produced a “possibility of a middle road” look at roadmapping a way ahead for Autonomous Weapons Systems. An opinion piece from Zachary Kallenborn on War on the Rocks look at What If the US Military Neglects AI? A paper in Nature provides an overview of open-ended evolution, as a part of artificial life. Gary Marcus and Ernest Davis publish a book on Rebooting AI: Building AI We Can Trust. The 57th Annual Meeting of the Association for Computational Linguistics occurred at the end of July, and Kate Koidan provides a summary of the top trends. The IEEE ranks robot creepiness with the top 100 creepy robots. Booz Allen releases a documentary on the Dawn of Generation AI. And the Naval Facilities Engineering and Expeditionary Warfare Center (NAVFAC EXWC) will host an industry day conference on cyber, control system, and machine learning in December.

 Click here to visit our website and explore the links mentioned in the episode. 

Welcome to Season 3.0! Andy and Dave discuss the AI in Advancement Advisory Council’s State of AI Advancement report, which takes a look at the impact of AI on roles within advancement. Researchers at Fudan and Changchun Institute of Optics announce a 500 MP camera (with associated cloud-powered AI) capable of identifying a face among tens of thousands. The U.S. National Science Foundation announces the National AI Research Institute, which anticipates approving $120M in grants next year. A recent solicitation from the Defense Innovation Unit seeks to understand trends in world events. And the JAIC has a new website. In research, OpenAI announces Dactyl, a robot hand capable of solving Rubik’s cube, as part of an effort to build a general purpose robot (transferring learning from simulation to the real world), and robust to perturbations such as broken fingers or intrusions by plush giraffes. Research accepted to ICLR 2020 demonstrates the application of deep learning to symbolic mathematics. Dan Gettinger of Bard College publishes The Drone Databook, cataloging the drones from 101 countries. The Carnegie Endowment for International Peace takes a look at the origins of AI Surveillance Technology in use around the globe. The Oliver Wyman Forum measures Global Cities’ AI Readiness, and Oxford Insights updates its Government AI Readiness Index. Arthur I Miller publishes the Artist in the Machine, while Marcus du Sautoy takes a look at The Creativity Code: Art and Innovation in the Age of AI. Lex Fridman and Gary Marcus have a discussion on AI. And Alexa will soon channel the voice of Samuel L Jackson.

Click here to visit our website and explore the links mentioned in the episode.