Last Week in AI #246: OpenAI fires CEO Sam Altman, UnitedHealth AI faces lawsuit over elderly care denial, DeepMind defines AGI, YouTube's AI music tools, and more!

OpenAI board fires CEO Sam Altman, UnitedHealth AI may have illegally denied elderly care, GPT-4V brings common sense to AVs, YouTube AI allows creators to clone singers' voices with permission

Top News

Silicon Valley Boardroom Coup Leads to Ouster of an AI Champion

In a shocking turn of events, OpenAI’s board has fired its co-founder and CEO, Sam Altman, following a series of disagreements over the safety and commercialization of AI. Altman reportedly clashed with board members, particularly Ilya Sutskever, the company's chief scientist, over the speed of development of generative AI, commercialization strategies, and measures to mitigate potential harm to the public. Following the announcement of Altman's dismissal, Greg Brockman, another co-founder of OpenAI, announced his departure from the company.

The board's decision to fire Altman sent shockwaves through the AI community and the wider tech industry. Microsoft CEO Satya Nadella, whose company is the largest investor in OpenAI, was reportedly blindsided and furious about the news. Mira Murati, the company’s CTO who helped develop some of OpenAI's key products, has been appointed as interim CEO. Meanwhile, Altman's attempts to raise billions of dollars from Middle Eastern sovereign wealth funds to create an AI chip startup to compete with Nvidia Corp had caused friction with board members who were concerned about the governance model of the new businesses.

The OpenAI board opened negotiations for Altman to return as CEO the day after the firing. By Monday morning the negotiations had broken down, and Microsoft said it was hiring both Sam Altman and Greg Brockman. They will lead an advanced research lab at Microsoft.

More coverage of this major story:

• 3 senior OpenAI researchers resign in the wake of Sam Altman's shock dismissal as CEO, report says - Three senior researchers have quit OpenAI following the dismissal of its CEO, Sam Altman.

• Details emerge of surprise board coup that ousted CEO Sam Altman at OpenAI - OpenAI fired CEO Sam Altman in a surprise move orchestrated by Chief Scientist Ilya Sutskever over concerns about the safety and speed of OpenAI's tech deployment, leading to the resignation of President Greg Brockman and three senior scientists, and reportedly making Microsoft CEO Satya Nadella furious.

• The Fear and Tension That Led to Sam Altman’s Ouster at OpenAI - Sam Altman, the former CEO of OpenAI, was ousted from his position due to tensions and concerns over the potential dangers of the company's AI technology and his perceived lack of attention to that risk, but there are discussions about his potential return.

• Microsoft Hires Sam Altman Hours After OpenAI Rejects His Return - Microsoft hires former OpenAI chief Sam Altman and president Greg Brockman to lead an advanced research lab, hours after OpenAI rejects Altman's return and names a new interim chief.

• OpenAI Staff Threaten to Quit Unless Board Resigns - OpenAI was in open revolt on Monday with 505 employees signing an open letter threatening to leave unless the board resigns and reinstates Sam Altman as CEO, along with cofounder and former president Greg Brockman. Altman was controversially fired by the board on Friday.

Lawsuit claims UnitedHealth AI wrongfully denies elderly extended care

A class action lawsuit has been filed against UnitedHealth Group Inc, alleging that the company's artificial intelligence algorithm wrongfully denies elderly patients' claims for extended care. The lawsuit claims that UnitedHealth uses the algorithm, known as nH Predict, to evaluate claims for post-acute care, such as stays in skilled nursing facilities and in-home care, and prematurely discontinues payment for healthcare services. Family members of two deceased UnitedHealth beneficiaries are seeking to represent a nationwide class of people on Medicare Advantage insurance plans, and claims for damages could reach billions of dollars. The lawsuit alleges that UnitedHealth's use of the algorithm violates contracts with patients and insurance laws, and seeks a court order to stop the practice and award money damages.

Google DeepMind wants to define what counts as artificial general intelligence

Google DeepMind has proposed a new definition for artificial general intelligence (AGI), outlining five ascending levels of AGI: emerging, competent, expert, virtuoso, and superhuman. The team believes that only the emerging level, which includes advanced chatbots like ChatGPT and Bard, has been achieved so far. The new definition aims to provide clarity and reduce confusion around the term AGI, which is often used inconsistently. The team's work has been welcomed by other AI researchers, who believe it will help to standardize discussions around AGI. The researchers, including DeepMind's co-founder Shane Legg and principal scientist Meredith Ringel Morris, published their paper online without any prior announcement.

YouTube previews AI tool that clones famous singers — with their permission

Google is introducing new AI features on YouTube titled Dream Track and Music AI Tools, allowing creators to develop tailor-made music tracks using just a text prompt or a hummed tune. The Dream Track feature, currently in testing with select US creators, can generate 30-second tracks imitating the styles of nine collaborating artists, including Charlie Puth and John Legend. The demonstration showed how text prompts can create tracks mimicking these artists with AI-generated lyrics, backing tracks, and voices in relevant styles. These AI tools are intended for use with YouTube's Shorts service. Other showcased Music AI Tools enable the creation of music tracks without traditional instruments. These tools are underpinned by Google's DeepMind's music generation model Lyria, which attaches an indiscernible SynthID watermark to each track to denote AI-generated audio.

Other News

Applications

Be My Eyes AI offers GPT-4-powered support for blind Microsoft customers - Microsoft is partnering with Be My Eyes to integrate OpenAI's ChatGPT-4 into its customer service for visually impaired users, allowing them to resolve technical issues and perform tasks without the need for human assistance.

Google will make fake AI products to help you find real gifts - Google's AI-powered Search Generative Experience (SGE) is introducing new shopping features that use AI to generate gift ideas and fashion items, as well as allowing users to generate photorealistic images of apparel they want and find similar-looking real products to purchase online.

AI tool helps ecologists monitor rare birds through their songs - Researchers have developed a deep learning AI tool called ECOGEN that generates lifelike bird sounds to enhance samples of underrepresented species, improving bird song classification accuracy by 12% on average and aiding in the monitoring of rare bird species.

Watch This Giant Chopstick Robot Handle Boxes With Ease - Robots are being developed to handle large, heavy items in warehouses, a task that is difficult for humans.

Google is embedding inaudible watermarks right into its AI generated music - Google is using SynthID, an inaudible watermarking tool, to identify AI-generated music created using its DeepMind AI Lyria model, allowing people to detect the origins of the music even after it has been compressed, altered, or had noise added.

Business

NVIDIA Supercharges Hopper, the World’s Leading AI Computing Platform - NVIDIA has introduced the NVIDIA HGX H200, a GPU with advanced memory that supercharges the world's leading AI computing platform, offering faster and larger memory for generative AI and high-performance computing workloads.

Microsoft unveils first AI chip, Maia 100, and Cobalt CPU - Microsoft has unveiled its first AI chip, the Azure Maia 100, which is optimized for generative AI tasks and is one of the largest chips on 5-nanometer process technology, along with the Azure Cobalt CPU, a microprocessor built in-house for cloud computing that achieves a 40% reduction in power consumption compared to other ARM-based chips.

Microsoft’s new toolkit makes running AI locally on Windows easier - Microsoft has unveiled Windows AI Studio, a toolkit that allows developers to fine-tune, customize, and deploy generative AI models for local, offline use in Windows apps, giving them the option to run models locally or in remote datacenters.

How AI helped Orangetheory’s legal team complete a 6-month project in half the time: ‘It’s straightforward math to see the cost savings’ - Orangetheory's legal team used AI to streamline and standardize user agreements across its franchises, reducing the time it took to review each contract by 80% and completing the project in half the expected time, resulting in significant cost savings and allowing the team to focus on customer needs and potentially boost revenue.

Korean AI startups Allganize, Neurophet raise over $35 mn - Korean AI startups Allganize and Neurophet have raised over $35 million in funding rounds, highlighting the increasing demand for AI technology in various industries.

Airbnb acquires AI startup for just under $200 million - Airbnb acquires AI startup Gameplanner.AI for just under $200 million, which will accelerate some of Airbnb's AI projects.

Meta launches AI-based video editing tools - Meta Platforms has launched two new AI-based video editing tools, Emu Video and Emu Edit, which allow users to generate short videos with captions and easily alter or edit videos with text prompts, as the company continues to make advancements in the AI space to compete with other tech giants.

Spotify to use Google's AI to tailor podcasts, audiobooks recommendations - Spotify is expanding its partnership with Google Cloud to use large language models (LLMs) in order to identify user listening patterns and provide tailor-made recommendations for podcasts and audiobooks.

Exclusive: Google in talks to invest in AI startup Character.AI -sources - Google is in talks to invest in Character.AI, an AI chatbot startup, to provide capital for training models and meet user demand, deepening their existing partnership.

Airbnb acquires AI startup for just under $200 million - Airbnb acquires AI startup Gameplanner.AI for just under $200 million, which will accelerate some of Airbnb's AI projects.

Amazon lays off hundreds in its Alexa division as it plows resources into AI - Amazon is cutting hundreds of jobs in its Alexa division as it shifts resources towards artificial intelligence initiatives.

Cruise co-founder and CEO Kyle Vogt resigns - Cruise co-founder and CEO Kyle Vogt has resigned from his position, following the suspension of Cruise's permits to operate self-driving vehicles on public roads after an incident involving a pedestrian.

Discord is already killing Clyde, its experimental OpenAI chatbot - Discord is shutting down its AI chatbot, Clyde, less than a year after its introduction, due to reasons that are not specified in the article.

Concerns

Why the Godfather of A.I. Fears What He’s Built - The Godfather of A.I. fears that digital intelligence may take over from biological intelligence due to its combination of immortality and replicability, and believes that current A.I. technology is talky and cerebral, stumbling at the borders of the physical.

AI Could Replace 4 Million Jobs in South Korea, BOK Study Says - Artificial intelligence could replace 4 million jobs in South Korea over the next two decades, with higher-income workers in analytic and cognitive roles being the most vulnerable, according to a study by the Bank of Korea.

Cruise Exec Omitted Pedestrian Dragging In Summary of Self-Driving Car Incident to California DMV, Email Shows - A senior executive at Cruise, a self-driving car company, failed to mention in a summary sent to the California Department of Motor Vehicles that one of their self-driving cars had dragged a pedestrian 20 feet, providing evidence that the company initially misled regulators about the incident.

White faces generated by AI are more convincing than photos, finds survey - White faces generated by AI are more convincing than photos, according to a survey, with people more likely to think AI-generated images are human than real photographs, raising concerns about identity theft and the potential for digital impostors.

Microsoft releases AI tool for photorealistic copying of faces and voices - Microsoft has released a new AI tool called Azure AI Speech that can generate photorealistic avatars and replicate voices, raising concerns about the creation of deepfakes.

Google fights scammers using Bard hype to spread malware - Google is suing scammers who are using the hype around generative AI to trick people into downloading malware by pretending to offer a paid version of Google's generative AI service Bard.

Feds Have No Idea How Many Times Cruise Driverless Cars Hit Pedestrians - The federal government is unaware of how many incidents there have been between Cruise autonomous vehicles and pedestrians in San Francisco in part because it does not have a good way to collect public complaints of unsafe autonomous vehicle driving, internal emails obtained by 404 Media show.

Underage Workers Are Training AI - Underage workers in countries like Pakistan and Kenya are bypassing age restrictions and joining online crowdsourcing platforms to perform data-labeling tasks for AI companies, highlighting the hidden human labor behind AI technology.

What the Doomsayers Get Wrong About Deepfakes - Deepfakes, manipulated videos created using AI, have raised concerns about the collapse of reality and the potential for widespread deception, but despite their proliferation, there is little evidence that deepfakes have had a significant impact or misled people in consequential ways.

Research

Generative AI Passes the Legal Ethics Exam in Study by LegalOn Technologies - Generative AI models, including OpenAI's GPT-4 and Anthropic's Claude 2, have been shown to pass the legal ethics exam, demonstrating their potential to assist lawyers in legal work and highlighting the advanced capabilities of large language models applied to law.

Google, Intel, Nvidia Battle in Generative AI Training - Google, Intel, and Nvidia are competing in the field of generative AI training, with MLPerf introducing a test for training large language models like GPT-3.

Race for First Drug Discovered by AI Nears Key Milestone - The race to discover the first drug using AI is approaching a significant milestone, with Insilico utilizing AI to accelerate the discovery process and reduce costs for a lung drug.

Better Machine Learning Models With Quantum Computers - Researchers at Terra Quantum have developed a method that combines classical and quantum computers to improve the training of machine learning models, resulting in better performance.

This AI robot chemist could make oxygen on Mars - Researchers in China have developed an AI-powered robot chemist that can extract oxygen from water on Mars, potentially eliminating the need for missions to carry oxygen from Earth.

One-2-3-45++: Fast Single Image to 3D Objects with Consistent Multi-View Generation and 3D Diffusion - A new AI model called One-2-3-45++ allows for fast conversion of single images into 3D objects with consistent multi-view generation and 3D diffusion.

On the Road with GPT-4V(ision): Early Explorations of Visual-Language Model on Autonomous Driving - Early explorations of GPT-4V(ision) as a visual-language model for autonomous driving.

Fine-tuning Language Models for Factuality - Fine-tuning language models using preference optimization algorithms and factuality rankings significantly improves their accuracy and reduces factual errors in open-ended generation settings.

in Wonderland: Large Multimodal Models for Zero-Shot Smartphone GUI Navigation - A new paper introduces MM-Navigator, an agent based on GPT-4V that can navigate and interact with smartphone interfaces using both text and image processing, showing promising results but also highlighting limitations in accuracy and zero-shot testing.

Meet JARVIS-1: Open-World Multi-Task Agents with Memory-Augmented Multimodal Language Models - Researchers have developed JARVIS-1, a multimodal agent designed for open-world tasks in Minecraft, which leverages pre-trained multimodal language models to interpret visual observations and human instructions, generating sophisticated plans for embodied control.

Instant3D: Instant Text-to-3D Generation - A new framework called Instant3D has been developed for fast text-to-3D generation, allowing for the creation of a 3D object from a text prompt in less than one second with a single run of a feedforward network.

This AI Paper Introduces RuLES: A New Machine Learning Framework for Assessing Rule-Adherence in Large Language Models Against Adversarial Attacks - A new machine learning framework called RULES is proposed to assess the rule-adherence of large language models (LLMs) against adversarial attacks, highlighting vulnerabilities in LLMs' behavior and calling for research to improve compliance and defend against attacks.

Mirasol3B: A Multimodal Autoregressive model for time-aligned and contextual modalities - The Librarian Bot found similar papers using the Semantic Scholar API and recommends giving a thumbs up if the comment was helpful.

Analysis

Fei-Fei Li Started an AI Revolution By Seeing Like an Algorithm - Fei-Fei Li, a key figure in the field of artificial intelligence, recounts her personal journey and the creation of ImageNet, highlighting the need for diversity in AI and the challenges of bias in machine learning.

Ghost, now OpenAI-backed, claims LLMs will overcome self-driving setbacks — but experts are skeptical - Ghost Autonomy, a startup backed by OpenAI, plans to explore the use of multimodal large language models (LLMs) in self-driving technology, despite skepticism from experts who argue that LLMs are not the right tool for autonomous driving and may not be efficient in solving the challenges of vehicular autonomy.

Policy

Amy Klobuchar and John Thune introduce legislation for creating generative AI framework - A bipartisan group of senators, led by Amy Klobuchar and John Thune, introduced legislation to establish guidelines and regulations for generative artificial intelligence, aiming to bring transparency, accountability, and security to its use.

OpenAI offers to pay for ChatGPT customers’ copyright lawsuits - OpenAI is offering to cover the legal costs of its ChatGPT customers facing copyright infringement suits, rather than removing copyrighted material from the chatbot's training dataset.

Music publishers ask court to halt AI company Anthropic's use of lyrics - Three music publishers are seeking a preliminary injunction to prevent AI company Anthropic from reproducing or distributing their copyrighted song lyrics, accusing the company of "systematic and widespread" infringement.

GM's Cruise suspends supervised and manual car trips, expands probes - GM's Cruise driverless car unit will pause all supervised and manual car trips in the U.S. and expand investigations following an accident, in an effort to rebuild public trust and undergo a full safety review.

---

Copyright © 2023 Skynet Today, All rights reserved.