Search Results

Your search for found 2049 results.

ai with ai: A Tesseract to Follow

/our-media/podcasts/ai-with-ai/season-3/3-37

In COVID-related AI news, Purdue University has built a website that tracks global response to social distancing, by pulling live footage and images from over 30,000 cameras in 100 countries. Simon Fong, Nilanjan Dey, and Jyotismita Chaki have published Artificial Intelligence for Coronavirus Outbreak, which examines AI’s contribution to combating COVID-19. Researchers at Harvard and Boston Children’s Hospital use a "regular" Bayesian model to identify COVID-19 hotspots over 14 days before they occur. In non-COVID AI news, the acting director of the JAIC announces a shift to enabling joint warfighting operations. The DoD Inspector General releases an Audit of Governance and Protection of DoD AI Data and Technology, which reveals a variety of gaps and weaknesses in AI governance across DoD. Detroit Police Chief James Craig reveals that the police department's experience with facial recognition technology resulted in misidentified people about 96% of the time. Over 1400 mathematicians sign and deliver a letter to the American Mathematical Society, urging researchers to stop working on predictive-policing algorithms. DARPA awards the Meritorious Public Service Medal to Professor Hava Siegelmann for her creation and research in the Lifelong Learning Machines Program. And Horace Barlow, one of the founders of modern visual neuroscience, passed away on 5 July at the age of 98. In research, Udrescu and Tegmark release AI Feynman 2.0, with unsupervised learning of equations of motion by viewing objects in raw and unlabeled video. Researchers at CSAIL, NVidia, and the University of Toronto created the Visual Causal Discovery Network, which learns to recognize underlying dependency structures for simulated fabrics, such as shirts, pants, and towels. In reports, the Montreal AI Ethics Institute publishes its State of AI Ethics. In the video of the week, Max Tegmark discusses the previously mentioned research on equations of motion and also discusses progress in symbolic regression. And GanBreeder upgrades to ArtBreeder, which can create realistic-looking images from paintings, cartoons, or just about anything.

ai with ai: Crime & Publishment

/our-media/podcasts/ai-with-ai/season-3/3-36

It’s a week of huge announcements! But first, in COVID-related AI news, Andy and Dave discuss a review paper in Chaos, Solitons, and Fractals that provides a more international focus on the role of AI and ML in COVID research. CSAIL teams with Ava Robotics to design a robot that maneuver between waypoints and disinfect surfaces of warehouses with UV-C light. C3.ai Digital Transformation Institute awards $5.4M to 26 AI researchers for projects related to COVID-19. In non-COVID news, the Association for Computing Machinery calls for the immediate suspension of facial recognition technologies until more mature and reliable. US lawmakers have introduced a bill that would ban police use of facial recognition, while separate bills seek to increase the AI talent available for the Department of Defense, and work to realign and rewire the JAIC within DOD. Over 2300 researchers sign a petition to Springer Nature to reject a publication from Harrisburg University, which developed facial recognition software to predict whether somebody was going to be a criminal. Meanwhile, researchers from Stanford demonstrate the problem of reproducibility by giving a data set of brain scans to 70 different researcher teams; no two teams chose the same workflow to analyze the data, and the final conclusions showed a sizeable variation. In a similar vein, researchers at Duke University examine the historical record of brain scan research and find poor correlation across experiments. In research, the "best paper" for the Conference on Computer Vision and Pattern Recognition goes to a team from Oxford, who use unsupervised learning methods and symmetry to convert single 2D images into 3D models. Researchers at Uber, the University of Toronto, and MIT use 3D simulated worlds to generate synthetic data for training LiDAR systems on self-driving vehicles. Calum MacKellar makes Cyborg Mind available, a look into the future of cyberneuroethics. And Johns Hopkins prepares for a second seminar on Operationalizing AI in Health.

ai with ai: Dust in the Mind

/our-media/podcasts/ai-with-ai/season-3/3-35

For COVID-related AI news, Andy and Dave discuss the Stanford Social Innovation Review report on the problem with COVD-19 AI solutions (e.g., data gaps, inconsistency, etc), and how to fix them. The National Endowment for Science Technology and the Arts (NESTA) provides a thorough report on AI and COVID-19, whose findings generally suggest that barriers might exist for the employment of AI in tackling COVID-19. In regular AI news, the US has its first known case of an erroneous arrest due to facial recognition technology, with the arrest of Robert Williams in Detroit in January 2020 (and disclosed on 24 June). The European Commission white paper on AI gets two more responses, from Facebook and from the Center for Data Innovation. Sergei Ivanov provides a breakdown of contributors for the upcoming International Conference on Machine Learning. Researchers have identified a new threat vector against neural networks, one that increases energy consumption and latency. And a follow-up with the Pulse upsampling tool shows a bias toward producing white faces, likely inherited from its training dataset, StyleGAN. In research, Denny Britz examines replicability issues in AI research, and how academic incentive systems are driving the AI research community toward certain types of research. The Marine Corps University Journal turns into the Journal of Advanced Military Studies, and its first issue focuses on innovation and future war. The Combat Studies Institute Press publishes On Strategy: A Primer, including a chapter on future war by Mick Ryan. And Major Nicholas Narbutovskih pens Dust, a story about two warring factions with different approaches to autonomous systems.

ai with ai: D/Generative

/our-media/podcasts/ai-with-ai/season-3/3-34

In COVID-related news, Nature publishes a review of COVID-19 AI tools, emphasizing that most tools are still in development and largely unproven. Inserm selects Expert System's AI support for its COVID-19 research and its group of over 10,000 researchers. Researchers provide in open-source a large annotated dataset of CT and X-ray images from COVID-19 patients, called the BIMCV COVID-19+. In regular AI news, Microsoft announces that it will not sell its facial recognition technology to police departments in the US until a national law is in place to help govern its use. On that note, a new federal bill in development, the Justice in Policing Act, contains policy guidelines on the use and limitations of facial recognition technology for police. OpenAI releases a commercial product API for accessing its AI models, to include the 175B parameter GPT-3, although other researchers are expressing concern over the lack of accountability on bias. Facebook announces the winner of its Deepfake Challenge, where the winning model achieved at 65% accuracy on a set of 10,000 previously unseen clips. And Boston Dynamics makes its robot dog, Spot, available for sale at $74,500 plus tax. In research, a team at Duke University introduces PULSE, which sharpens blurry images, in essence by exploring the space of plausible high-res images that could result in the blurry image. The report of the week comes from Perry World House, who published the results of a Policy Roundtable on AI hosted last fall. The Stockholm International Peace Research Institute and the International Committee of the Red Cross offer their take on Limits on Autonomy in Weapon Systems, by identifying the practical elements of human control. The review of the week from University of Waterloo provides an overview of text detection and recognition in the wild. MacroPolo provides a snapshot of Global AI Talent, using participants from the 2019 NeurIPS. Spring-Verlag provides yet another free text, from Eiben and Smith, on an Introduction to Evolutionary Computing. And NavyCon 2020 provides brief snapshots on "navies, science fiction, and great power competition" from a host of participants.

ai with ai: Oura-boros

/our-media/podcasts/ai-with-ai/season-3/3-33

In COVID-related AI news, Andy and Dave discuss an announcement from WVU Rockefeller Neuroscience Institute, WVU Medicine, and Oura Health, with the ability to predict COVID-19 related symptoms up to three days in advance via biometric monitoring. Japan's M3 is teaming with Alibaba's AI Tech to provide CT-scan capability to hospitals that can identify COVID-related pneumonia. The Pentagon taps into the virus-relief CARES Act to use AI for virus cure and vaccine efforts. Rockefeller announces efforts to use GPT-2 to automatically summarize COVID-19 medical research articles, but the results aren’t that great. In regular AI news, IBM announces it is no longer offering general-purpose facial recognition or analysis software, due to concerns about the technology being used to promote racism. And in a related announcement, Amazon places a one-year moratorium on allowing law enforcement to use its Rekognition facial recognition platform. USSOCOM has posted an RFI for potential contractors to provide its Global Analytics Platform, a $300-600M contract that would follow its previous eMAPS contract. And NASA launches its Entrepreneurs Challenge, seeking new ideas for space exploration. In research, from the University of Pennsylvania, UC Berkeley, Google Brain, University of Toronto, Carnegie Mellon University, and Facebook AI, comes a different approach to defining intrinsic motivation for taskless problems, wherein agents seek out future inputs that are expected to be novel. The report of the week comes from the Stanley Center for Peace and Security, with a look at The Militarization of AI. Researchers at Beijing Academy and Cambridge University come together to pen a white paper calling for "cross-cultural cooperation" on AI ethics and governance. Efron, Hastie, and Cambridge University Press provide Computer Age Statistical Inference for free. And DeepMind and the UCL Centre for AI are producing a Deep Learning Lecture Series.

ai with ai: The GPT Blob

/our-media/podcasts/ai-with-ai/season-3/3-32

In this week's COVID-related AI news, Andy and Dave discuss "SciFact" from the Allen Institute for AI, which built on neural network VeriSci and can link to supporting or refuting materials for claims about COVID-19. Berkeley Labs releases COVIDScholar, which uses natural language processing text-mining to search over 60,000 papers and draw insights and connections. Berekely Labs also announces plans to use machine learning to estimate COVID-19's seasonal cycle. In non-COVID AI news, Google publishes a response to the European Commission's white paper on AI, cautioning that their definition of AI is far too broad and risks stifling innovation. CSET maps where AI talent is produced in the U.S., where it gets concentrated, and where AI funding equity goes. In research, OpenAI releases GPT-3, a 175B parameter NLP model, and shows that massively scaling up the language model greatly improves task-agnostic few-shot performance. A report from the European Parliament's Panel for the Future of Science and Technology shows the ethics initiatives of nations around the globe. A review paper in Science suggests that progress in AI has stalled (perhaps as much as 10 years) in some fields. Abbass, Scholz, and Reid publish Foundations of Trusted Autonomy, a collection of essays and reports on trustworthiness and autonomy. And in the video of the week, CSIS sponsored a conversation with (now retired) JAIC Director, Lt Gen Shanahan.

ai with ai: Eye, Pac-Bot

/our-media/podcasts/ai-with-ai/season-3/3-31

In COVID-related AI news, Andy and Dave discuss work from Mount Sinai researchers, who have created an AI system that uses CT scans to diagnose patients with COVID-19. MIT and IBM Watson announce plans to fund 10 AI research projects to find COVID-19. The National Security Commission on AI releases its second white paper on COVID-19, on mitigating economic impacts of the COVID-19 pandemic and preserving US strategic competitiveness in AI. In non-COVID AI news, DARPA’s Gamebreaker project holds a virtual kickoff meeting of its program, seeking to model and then break the game balance. The United Nations Secretary-General releases a report on the protection of civilians in armed conflict. And the JAIC unveils its "business process transformation" initiative. In research, Hong Kong University of S&T publishes research on EC-Eye, an artificial eye that "sees" like a human eye. Other research demonstrates that neural networks trained for prediction mimic the diverse features of biological neurons and perception. And NVidia, University of Toronto, Vector Institute, and MIT publish GameGAN, and generative model that learns visually to imitate a desired video game (in this case, observing and replicating the gameplay of Pac-Man). The report of the week comes from NATO, which publishes a look at S&T Trends 2020-2040. Wolfgang Ertel pens the book of the week, with Introduction to AI (2nd Edition), free through Springer. The University of Southern California hosts a virtual symposium AI for COVID-19 in LA. And a collaboration between Google and the Getty Museum produces Art Transfer, transforming photos using the style of different artists.

ai with ai: Afterburn-in

/our-media/podcasts/ai-with-ai/season-3/3-30

In COVID-19-related AI news, Andy and Dave discuss the Novel Coronavirus Research Compendium, a curated resource from Kate Grabowski, an epidemiologist at Johns Hopkins, with the goal of providing a smaller, but higher-quality, data set on coronavirus research. Primer AI uses natural language processing to summarize the latest information on COVID-19. C3.ai provides a COVID-19 data lake with accompanying knowledge graphs, and is ready for use with R or Python. On 1 June, the Stanford Institute for Human-Centered AI will host a free virtual conference on the way ahead for AI as the world recovers from COVID-19. And Singapore hires Boston Dynamics’ Spot to roam a public park (controlled by a park ranger) and play a recorded message to encourage social distancing. In non-COVID news, Sony unveils the first line of cameras with a built-in image classifying AI. Researchers at UCLA and Baylor demonstrate the ability to dynamically stimulate the brain cortex to mirror the motion of writing. Booz Allen wins an $800M contract to support the Joint AI Center with AI services. Thomas Dimson uses GPT-2 to create and define words that don’t exist. And August Cole - best-selling author, lecturer, and consultant on national security issues - joins for a discussion on fiction intelligence (FICINT), the role that it plays in thinking about possible futures, and how those ideas play out in his latest book, Burn-In.

ai with ai: HurriCOVID Season

/our-media/podcasts/ai-with-ai/season-3/3-29

In COVID-related AI news, Andy and Dave discuss an approach from FiveThirtyEight that uses a mini-model-ensemble to predict possible trajectories for the COVID-19 death toll. MIT Tech Review has released a tracker for COVID-19 tracing trackers, which includes information on how they work and what policies they have in place. In non-COVID-related AI news, DIU releases a solicitation for Vigilante Keeper, an AI solution for detecting behavioral changes that might indicate increased vulnerability. OpenAI releases an analysis that shows the amount of computation needed to train an ImageNet classifier decreases by a factor of 2 every 16 months, which suggests algorithm progress has resulted in more gains than increased hardware efficiency. The Library of Congress is using machine learning to digitize and organize photos from old newspapers. Microsoft unveils a new tool in Word that makes sentence-level suggestions. And MIT Tech Review Insights publishes an examination of Asia’s advantage in AI with a look at the Asia-Pacific region in the Global AI Agenda. In research, Andy and Dave discuss RTFM (Read to Fight Monsters) from Facebook, which uses roguelike procedural generation to dynamically create goals, monsters, and other attributes, which agents then attempt to fight. The book of the week comes from Miroslav Kubat, with the second edition of An Introduction to Machine Learning. The Australian Defense College has announced the winners of its 2020 Sci-Fi Writing Competition. The full documentary for AlphaGo – The Movie, is now available on YouTube. The proceedings are now available from a federal health virtual forum on AI for COVID-19 Response. CSET will host a discussion on lessons learned for Algorithmic Warfare in DoD on 27 May. And LessWrong by Stuart Armstrong takes a look at Kurzweil's predictions (from 1999) about 2019.

ai with ai: Finding Lenia

/our-media/podcasts/ai-with-ai/season-3/3-28

Andy and Dave discuss a white paper from the National Security Commission on AI, on Privacy and Ethics Recommendations for Computing Applications Developed to Mitigate COVID-19. The Office of the Undersecretary of State for Arms Control and International Security issues a second point paper on lethal autonomous weapons systems, with AI, Human-Machine Interaction, and Autonomous Weapons. DARPA announces its Air Space Total Awareness for Rapid Tactical Execution (ASTARTE) program, which aims to use low-cost sensors to create a better common operating picture. The Joint AI Center establishes a Data Governance Council to establish an enterprise-wide data governance framework. The JAIC also releases an AI Primer for DoD officials. The U.S. Patent and Trademark office denies patents on the behalf of AI systems. And Google Health describes the challenges in transitioning to clinical environments a system designed to detect diabetic eye disease. In research from the University of Bordeaux, researchers demonstrate the ability to give algorithms intrinsically-motivated goal exploration to enable them to search out interesting patterns in Lenia, an analog version of Conway's Game of Life. A review paper provides an overview of how neural networks sometimes attempt to "short circuit" learning. Peters, Janzing, and Schölkopf and MIT Press make Elements of Causal Inference available. The International Conference on Learning Representation makes its 2020 session available through a slick interface, covering the nearly 700 papers. And OpenAI releases Jukebox, an attempt to create music of a specified style, when given lyrics.