Discover more from Last Week in AI
Last Week in AI #195: Meta's AI can negotiate with and persuade people in the game of Diplomacy, how AI is used by the World Cup, AI-powered insect trap, and more!
Meta's CICERO achieves human-level performance in the strategy game Diplomacy, World Cup uses AI to track soccer balls and assist referees, AI-powered insect trap reduce need for pesticides
Meta unveiled CICERO, an AI agent that achieves human-level performance at the game of Diplomacy. Unlike board games such as Go and Chess, Diplomacy is much more complicated and requires players to negotiate, persuade, and collaborate with other players through language. At every turn, players need to send messages to one another privately, and at the end of the turn, everyone reveals their actions. CICERO has two components - a planning engine that plans its overall strategy and predicts the moves other players will make, and a dialogue engine that generates the text messages to be sent to other players in order to negotiate and form collaborations. In online games, CICERO “achieved more than double the average score of the human players and ranked in the top 10% of participants who played more than one game.”
Our take: This is a breakthrough in the same order as Deep Blue and AlphaGo, and it convincingly demonstrates a few capabilities that prior AI systems lack, like controllable dialogue, inferring human intentions, and facilitating meaningful interactions with humans over time. Meta AI was clear that this research isn’t about the game of CICERO, but rather about eventually enabling more grounded applications that require human-AI interactions. It’s also easy to imagine how a system like this could be exploited by malicious actors. Meta did open-source the models and algorithms behind CICERO, which should allow other researchers to both build on top of this system as well as to study ways to address potential harms.
The 2022 FIFA World Cup in Qatar is in progress. A number of worries accompany a large sporting event such as this one: botched calls, large crowds, overheating, public safety, and more. AI technology, which has been deployed in sports scenarios before, is again being deployed to mitigate a number of potential issues. Among the innovations being used are a soccer ball equipped with motion sensors to report precise location data, algorithmic video assistant referees that will help on-field referees make accurate calls, and more than 15,000 cameras to track people's movements throughout the games.
Our Take: This isn't the first case of AI being applied to sports. I think removing some of the obvious cases of human error--bad referee calls being an important one--is a positive advance for the world of sports. I expect also that analysis done by AI systems will enable athletes and coaches to better understand their strengths and weaknesses and train better. But I wonder how the entry of AI systems into sports will affect the games more broadly--I think that the changes will not be analogous to chess, for instance, since sports are not purely mental and robots aren't that great just yet (and if they were, most would probably agree that a human team vs a robot team would be an unfair matchup). While these systems will likely be helpful, I do feel there is also a physical "intuition" about sports that comes from years of practice and mistakes that data might help direct and speed up, but won't replace.
According to the UN, pests destroy up to 40% of crops worldwide each year, leading to $220 billion in economic losses. Trapview is a company that aims to tackle this issue by developing devices that trap and identify pests. Each device uses pheromones to attract pests. It then photographs the insect, identifies its species using AI, and alerts the farmers of any potential impact that needs to be addressed. This helps to reduce the number of times farmers need to visit the field, and it also allows more targeted applications of pesticides, significantly reducing the use of chemical sprays. This more proactive approach to pest management stands in contrast to more reactive approaches in the past, and the company has sold more than 7.5k devices in 50 countries since its launch.
Our take: This is another example of an impactful, although perhaps not as glamorous, application of AI. It’s also a sign of how mature computer vision techniques have become just in recent years, where companies not focused on AI can also leverage these advances in ways that make economic sense. This makes me more excited about AI’s role in agriculture and how it can help us to grow more food in a more sustainable fashion.
CMU, Berkeley Researchers Design System Creating Robust Legged Robot - "A Low-Cost Robot Ready for any Obstacle”
Low-Cost Robot Navigates Nearly Any Obstacle - "A team of researchers has designed a robotic system that enables a low-cost, small legged robot to navigate nearly any obstacle or terrain. The robot can climb and descend stairs nearly its height or navigate rocky, slippery, uneven, steep and varied terrain."
3D for everyone? Nvidia’s Magic3D can generate 3D models from text - "On Friday, researchers from Nvidia announced Magic3D, an AI model that can generate 3D models from text descriptions. After entering a prompt such as, "A blue poison-dart frog sitting on a water lily," Magic3D generates a 3D mesh model, complete with colored texture, in about 40 minutes."
Will artificial intelligence ever discover new laws of physics? - "Algorithms can pore over astrophysical data to identify underlying equations. Now, physicists are trying to figure out how to imbue these “machine theorists” with the ability to find deeper laws of nature"
Self-organization: What robotics can learn from amoebae - "LMU researchers have developed a new model to describe how biological or technical systems form complex structures without external guidance. Amoebae are single-cell organisms."
Teaching photonic chips to 'learn' - "A multi-institution research team has developed an optical chip that can train machine learning hardware."
Massive traffic experiment pits machine learning against ‘phantom’ jams - "Many traffic jams are caused by human behavior: a slight tap on the brakes can ripple through a line of cars, triggering a slowdown — or complete gridlock — for no apparent reason."
Powerful rare-earth free magnet ‘evolved’ and refined by machine learning algorithm - "A rare-earth free magnetic material with similar properties to the rare-earth magnets found in everything from wind turbines to computer hard drives has been discovered by US researchers using a machine learning-guided approach."
Using machine learning to infer rules for designing complex mechanical metamaterials - "Two combinatorial mechanical metamaterials designed in such a way that the letters M and L bulge out in the front when being squeezed between two plates (top and bottom). Designing novel metamaterials such as this is made easy by AI."
A bot that watched 70,000 hours of Minecraft could unlock AI’s next big thing - "OpenAI has built the best Minecraft-playing bot yet by making it watch 70,000 hours of video of people playing the popular computer game."
Running Fast Transformers on CPUs: Intel Approach Achieves Significant Speed Ups and SOTA Performance - "Large transformer language models (LM) that scale up to billions of parameters have demonstrated state-of-the-art performance across a wide variety of natural language processing (NLP) tasks."
Real-Life ‘Invisibility Cloak’ Stops AI Cameras From Recognizing People - "Scientists have developed a real-life “invisibility cloak” that tricks artificial intelligence (AI) cameras and stops them from recognizing people."
BBC documentary used face-swapping AI to hide protesters' identities - "An artificial intelligence helped to protect the identity of people who took part in a violent anti-government protest in Hong Kong by swapping their faces for those of actors."
More self-driving cars are riding into SF - "Expect to see more ghost-riders, more commonly called self-driving vehicles, on the streets of San Francisco."
Copilot Lawyers Checking Claims Against Other AI Companies - "The attorneys that filed the GitHub Copilot lawsuit say they’re getting messages every day from creators concerned about how their work is being used by artificial intelligence companies, according to lead attorney Matthew Butterick."
Microsoft Registers New Patent For AI-Generated Audio in Games and Movies - "Microsoft has registered a new patent on the use of AI-generated music/soundtracks/audio in an extensive range of media, including movies, video games, live recordings and related fields."
Exclusive: Care.ai gets $27M for hospital sensors - ""I want to kill the EMR," says Care.ai CEO Chakri Toleti."
Amazon Backed Out of Taking a Stake in Argo. Then the Self-Driving Startup Folded. - "Amazon.com Inc. emerged as a potential savior for Argo AI, the now-defunct startup backed by two of the world’s biggest automakers, before the deal fell apart because of a sputtering economy, concerns about control and flagging faith in fully autonomous driving."
Picsart AI Image Generator Being Used to Make a Million Images a Day - "In just 20 days since it was launched, Picsart users are creating more than a million images a day with the company’s text-to-image generator. The company will begin rolling out the feature to Android users today."
This Copyright Lawsuit Could Shape the Future of Generative AI - "The tech industry might be reeling from a wave of layoffs, a dramatic crypto-crash, and ongoing turmoil at Twitter, but despite those clouds some investors and entrepreneurs are already eyeing a new boom—built on artificial intelligence that can generate coherent text, captivating images, a"
Harvey, which uses AI to answer legal questions, lands cash from OpenAI - "Harvey, a startup building what it describes as a “copilot for lawyers,” today emerged from stealth with $5 million in funding led by the OpenAI Startup Fund, the tranche through which OpenAI and its partners are investing in early-stage AI companies tackling major problems."
Lawsuit Takes Aim at the Way A.I. Is Built - "A programmer is suing Microsoft, GitHub and OpenAI over artificial intelligence technology that generates its own computer code."
New Cerebras Wafer-Scale ‘Andromeda’ Supercomputer Has 13.5 Million Cores - "Cerebras unveiled its new AI supercomputer Andromeda at SC22. With 13.5 million cores across 16 Cerebras CS-2 systems, Andromeda boasts an exaflop of AI compute and 120 petaflops of dense compute"
Google has a secret new project that is teaching artificial intelligence to write and fix code. It could reduce the need for human engineers in the future. - "Google is working on a secretive project that uses machine learning to train code to write, fix, and update itself. This project is part of a broader push by Google into so-called generative artificial intelligence, which uses algorithms to create images, videos, code, and more."
AI giant Baidu shrugs off US chip export restrictions as having 'little impact' - "Says sanctions could even accelerate China's drive for silicon self-sufficiency"
Stable Diffusion update removes ability to copy artist styles or make NSFW works - "Stable Diffusion, the AI that can generate images from text in an astonishingly realistic way, has been updated with a bunch of new features."
Is AI art really art? This California gallery says yes - "As artificial intelligence becomes increasingly popular for generating images, a question has roiled the art world: Can AI create art? At bitforms gallery in San Francisco, the answer is yes."
US FDA, Health Canada, and UK MHRA have Jointly Identified 10 Guiding Principles that can Inform the Development of Good Machine Learning Practice (GMLP) - "The UK’s MHRA, The U.S. Food and Drug Administration (FDA), and Health Canada have come together to work on the ten basic guiding principles to develop Good Machine Learning Practice (GMLP)."
What does Meta AI’s Diplomacy-winning Cicero Mean for AI? - "Cicero the human being (106-43 BC) was a celebrated politician, an orator, and a writer; his historical and philosophical importance is still debated two thousand years later."
How AI Will Completely Dominate the Animation Industry In Less Than 5 Years - "If you're looking to get into animation as a career, you have less than five years. If Toei Animation (the studio in charge of One Piece and DBZ) is already doing it... you should probably take it seriously."
AI Has Run 100,000 Simulations And Predicted The 2022 World Cup Winner - "It's not looking great for the USA."
Copyright © 2022 Skynet Today, All rights reserved.