Last Week in AI #188: Google's Imagen Text-to-Video Model, slow progress in self-driving cars, AI's role in chess scandal, and more!

Google showcases high-res text-to-video with Imagen, self-driving cars making progress but very slowly, AI chess engines in chess scandals might be sign of things to come

Top News

Google’s newest AI generator creates HD video from text prompts

Not to be outdone by Meta’s recent make-a-video, Google announced its Imagen Video AI model, which also generates video clips from text descriptions. Compared to other text-to-video models, Imagen Video results have much higher resolutions and seem to inherit Imagen’s (text-to-image model) ability to parse precise descriptions and generate coherent texts. The model was trained on 14 million video-text pairs and 60 million image-text pairs, highlighting the benefits of training on large-scale datasets. Citing potential harms and misses, Google does not have plans to release the model in the near future.

Our take: The speed of progress in generative models is truly breathtaking. Google actually released another text-to-video model, Phenaki, that is focused on long-horizon video generation instead of high-resolution. The company is working on combining the two models to get the best of both worlds. While Google does not intend to release the model, it would be surprising if similar models are not released on the Internet by organizations like Stability.ai in the next couple to a few months. Given the Cambrian explosion of text-to-image applications following only a few weeks of stable diffusion’s release, one can only imagine what the open-source community can do with a text-to-video model. It is certainly an exciting time for generative AI.

Even After $100 Billion, Self-Driving Cars Are Going Nowhere

Self-driving cars promise a radically different future. Over the past few years, companies developing autonomous vehicles have offered ambitious hopes: safer roads, reduced traffic congestion, recaptured commute time, and greater independence for those unable to drive themselves. It sounds great, but the companies involved--Google, GM, Ford, Tesla, and Zoox, among others--are still far from fulfilling their promise of vehicles that can, all by themselves, navigate through a panoply of situations with no human input or oversight. As Bloomberg reports in great detail, our driverless future is continuing to look farther and farther away--the losses are getting bigger, and prominent detractors are getting louder.

Our Take: It's very easy to look at problems like this in retrospect and say "well, of course that was going to be more difficult than we thought." For many brilliant technologists, the self-driving car did seem just around the corner: perhaps there is something about being embedded in such a rapidly-moving field that distorts our depth perception when we stare at hard problems. But here we are, 100 billion dollars and many accidents and failures later. The future where self-driving vehicles are capable of truly autonomous navigation does seem incredibly far in the future, and perhaps many hundreds of billions more will be poured down the drain as companies continue to struggle with the challenge. I'm hopeful that at the very least, we'll make important improvements and solve problems along the way that make our lives better, even if the holy grail seems to fade further into the distance as we continue the pursuit.

AI’s Grandmaster Status Overshadows Chess Scandal

Magnus Carlsen, the world chess champion, accused Hans Niemann, a U.S. grandmaster, of cheating during their game at the Sinquefield Cup, in St. Louis, Mo. A more detailed report from Chess.com suggests that Niemann has cheated in more than 100 games because his moves in those games aligned too well with those that would’ve been chosen by computer chess engines. Cheating scandals aren’t new to chess, but what makes this one stick is how it highlights humans’ “perpetual underdog role” in chess. The best engine is rated 678 points ahead of Magnus, a gap that predicts a 99% win rate, and these engines are efficient enough to run on smartphones.

Our Take: The dynamic of chess AI and the chess community is worth paying attention to because it might be a sign of things to come as more and more human activities get “out-competed” by their AI counterparts. Chess engines make for great sparring partners to play against and lower the age bar of chess prodigies, but they also make cheating much easier to perform. The mass proliferation of future AI technologies will likely also have such mixed implications, and it will be up to human users to set up rules and norms to ensure ethical use.

Other News

Research

DeepMind unveils first AI to discover faster matrix multiplication algorithms - "DeepMind unveiled AlphaTensor, the “first artificial intelligence system for discovering novel, efficient and provably correct algorithms.”"

A Machine Learning View of Supernova GWs - "The authors of today’s paper use machine learning to classify the progenitors of over 400 supernova waveforms and determine what qualities, if any, best predict the original mass of the star that died."

New algorithms help four-legged robots run in the wild - "A team led by the University of California San Diego has developed a new system of algorithms that enables four-legged robots to walk and run on challenging terrain while avoiding both static and moving obstacles."

New Approach to Machine Learning Could Make Chaos More Predictable - "The vast number-crunching capabilities of artificial intelligence systems mean we can better predict the future of chaotic systems based on fewer and fewer patterns of the past – and a new algorithm is adding even more accuracy to the process."

German Researchers Improve Climate Models with Machine Learning - "GANs Can Reduce Electricity Consumed in Climate Modelling"

Machine learning and deep learning applications in microbiome research - "The many microbial communities around us form interactive and dynamic ecosystems called microbiomes. Though concealed from the naked eye, microbiomes govern and influence macroscopic systems including human health, plant resilience, and biogeochemical cycling."

Google’s new AI can hear a snippet of song—and then keep on playing - "AudioLM, developed by Google researchers, generates audio that fits the style of the prompt, including complex sounds like piano music, or people speaking, in a way that is almost indistinguishable from the original recording."

A New Study by Google and DeepMind Introduces Geometric Complexity (GC) for Neural Network Analysis and Understanding of Deep Learning Models - "Through theoretical and empirical methods, the researchers show that a wide variety of regularisation and training heuristics can regulate geometric complexity using a variety of mechanisms."

Algorithms predict sports teams' moves with 80% accuracy - "Algorithms developed in Cornell's Laboratory for Intelligent Systems and Controls can predict the in-game actions of volleyball players with more than 80% accuracy, and now the lab is collaborating with the Big Red hockey team to expand the research project's applications."

Applications

What Makes a Champagne Vintage Great? Ask a Deep Learning Model - "Bollinger is data mining its meticulous records to predict vintage years and weather the challenges of climate change."

AI music generators could be a boon for artists — but also problematic - "Stability AI, the company behind Stable Diffusion, is tackling music"

AI text-to-image generators bring delight : and concern - "The number of artificial intelligence text-to-image generators is increasing rapidly, giving the public an array of new tools to create digital imagery."

Business

Amazon Abandons Autonomous Home-Delivery Robot in Latest Cut - "Amazon.com Inc. is abandoning its home delivery robot, the latest sign that the e-commerce giant is starting to wind down experimental projects amid slowing sales growth."

Uber is back in the robotaxi game after signing 10-year deal with Motional - "Nearly two years after selling its autonomous vehicle research division to a competitor, Uber is finally back in the robotaxi business."

New report documents the business benefits of ‘responsible AI’ - "Firms are falling short in developing AI systems that are good for individuals and society. Here are three best practices for getting it right. "

Want fries with that? Robot makes French fries faster, better than humans do - "Fast-food French fries and onion rings are going high-tech, thanks to a company in Southern California. Miso Robotics Inc in Pasadena has started rolling out its Flippy 2 robot, which automates the process of deep frying potatoes, onions and other foods."

Pizza robot truck startup Stellar snags $16.5M from Jay-Z - "Dive Brief: Stellar Pizza, a company that makes trucks equipped with robotic pizza makers, has raised $16.5 million in a funding round backed by Jay-Z’s Marcy Venture Partners, multiple outlets reported."

Artificial Intelligence (AI) in Medical Diagnostics Market worth $5.5 billion by 2027 - Exclusive Report by MarketsandMarkets™ - "AI in Medical Diagnostics Market is projected to grow from USD 1.0 billion in 2022 to USD 5.5 billion by 2027, at a CAGR of 39.9% from 2022 to 2027, according to a new report by MarketsandMarkets™."

RBI Plans To Extensively Use AI, Machine Learning To Improve Regulatory Supervision - "The Reserve Bank is planning to extensively use advanced analytics, artificial intelligence and machine learning to analyse its huge database and improve regulatory supervision on banks and NBFCs."

Lessons Learned: The Journey to Real-Time Machine Learning at Instacart - "Instacart incorporates machine learning extensively to improve the quality of experience for all actors in our “four-sided marketplace” "

Reddit Acquires Oterlu, Developer of Machine-Learning-Powered Moderation Tools - "The four-person team will join the platform’s safety team."

Seattle startup Xembly raises $15M for AI ‘chief of staff’ that automates tedious conversational tasks - "Seattle-based Xembly, makers of an artificial intelligence “chief of staff” designed to automate tedious tasks such as meeting notes, scheduling, to-do lists, and more, has raised $15 million in new funding."

US DoD Extends AI, Machine Learning Contract With Palantir Technologies - "The US Army Research Laboratory has extended its contract with Palantir Technologies to continue advanced artificial intelligence (AI) and machine learning (ML) services for the US Department of Defense."

These 5-foot-1 security robots made by Cobalt are patrolling workplace hallways. Here's how companies are using them to protect offices and employees. - "While not quite as menacing, the robotics startup Cobalt makes 5-foot-1 robots meant to replace human security guards. They patrol offices and report any observed anomalies."

Exclusive: Boston Dynamics pledges not to weaponize its robots - "The company joined other robotics firms in pledging to curtail projects that turn robots into war machines."

AI-powered supply chain visibility platform Altana bags $100M - "Trade wars, military conflicts, the rise of e-commerce and sustainability concerns are driving changes to supply chain networks and trade flows. Seventy-four percent of supply chain companies made changes to their networks within the past two years, according to Gartner."

Gather AI secures new cash to scan inventory in warehouses using drones - "Gather AI, a startup using drones to inventory items in warehouses, today announced that it raised $10 million in a Series A round "

Google's Tensor G2 Chip Is Key to Many Pixel 7 AI-Powered Abilities - "Google on Thursday revealed its Tensor G2 smartphone processor, the second-generation design of the digital brains that accelerate AI abilities on its Pixel 7 and Pixel 7 Pro smartphones."

UK AI investment reaches record $3.6bn despite economic downturn - "UK artificial intelligence (AI) startups have attracted a record $3.6bn in investment so far this year in a sign that the sector remains strong despite the economic downturn."

Analysis

Artificial Intelligence & Machine Learning Report - "VC funding slid 27.8% quarter-over-quarter for AI & ML startups in Q2. Of the 70 product categories tracked by PitchBook, only 21 are on pace to grow in VC funding in 2022, driven by leading vertical applications including sales & marketing, information security, and drug discovery."

In the Battle With Robots, Human Workers Are Winning - "Why do I still have a job? It’s a question readers ask me often, but I mean it more universally: Why do so many of us still have jobs?"

Policy

AI Blueprint Could Handcuff America’s Ability to Compete on the Global Stage - "Jordan Crenshaw, Vice President of the U.S. Chamber’s Technology Engagement Center, issued the following statement today in response to the White House Office of Science and Technology Policy’s Blueprint on Artificial Intelligence."

White House unveils ‘AI bill of rights’ as ‘call to action’ to rein in tool - "The non-binding guidelines call on companies to curb uses that lead to harm."

Weaponized Robots Letter Calls for Policy, Tech Fixes - "Manufacturers’ initiative pledges no militarized robots for nonmilitary purposes"

Expert Opinions

Interview: Why Mastering Language Is So Difficult for AI - "The field of artificial intelligence has never lacked for hype. Back in 1965, AI pioneer Herb Simon declared, “Machines will be capable, within 20 years, of doing any work a man can do."

---

Copyright © 2022 Skynet Today, All rights reserved.