The epilogue to the OpenAI drama - Weekly Roundup - Issue #443
Plus: DeepMind's AI discovers millions of new materials; $101M XPRIZE Healthspan competition; Italy bans lab-grown meat; a boxing robot; and more!
Welcome to Weekly Roundup Issue #443. This week brings a resolution at OpenAI, with Sam Altman officially resuming his role as CEO. In other developments, the XPRIZE Foundation has launched a $101 million competition to enhance healthspan. Additionally, DeepMind’s AI has made significant strides by identifying thousands of novel materials. Meanwhile, Italy has imposed a ban on lab-grown meat, the Pentagon has announced a new drone program to compete with China, Amazon launched an AI assistant for AWS, and researchers have unveiled a boxing robot.
As Sam Altman returns as the CEO of OpenAI, it appears the drama at the company is coming to an end. In a post published on November 29th, OpenAI confirmed Sam Altman's reinstatement as CEO. Mira Murati, who briefly served as interim CEO, has resumed her prior role as CTO. Greg Brockman, having left the company in protest of Altman's firing, returns as President. Everyone who followed Brockman's lead and left the company has now rejoined OpenAI.
However, in light of recent events at OpenAI, there will be changes to how OpenAI is organised. The first major change involves the board of directors. Last week, OpenAI announced a new initial board, with former Salesforce co-CEO Bret Taylor as chair and ex-U.S. Treasury Secretary Larry Summers. Quora CEO Adam D'Angelo, part of the board that ousted Altman, will remain on the board. Ilya Sutskever, chief scientist at OpenAI and the instigator of the coup, is no longer on the board. The previous board, apart from Altman, Brockman, Sutskever, and D'Angelo, included entrepreneur Tasha McCauley and Helen Toner, director of strategy at Georgetown's Center for Security and Emerging Technology.
Many expected Microsoft, OpenAI’s largest investor and key technological and business partner, to gain a seat on the new board, especially following Satya Nadella’s comment that Microsoft will 'never going to get back into a situation where we get surprised like this.' According to Reuters, Microsoft will indeed have a board seat but as a non-voting observer. This role allows Microsoft’s representative to attend OpenAI's board meetings and access confidential information but does not grant voting rights on decisions, including the election or selection of directors.
Another aspect set for change is OpenAI’s governance structure. Currently, OpenAI maintains a unique blend of for-profit and non-profit entities, with the non-profit entity owning and overseeing the for-profit one. In an interview with The Verge, Altman acknowledged that 'clearly our governance structure had a problem.' When pressed for details, he remarked that it is too early to elaborate. The board will conduct a review of the recent events, and their findings will guide future decisions. However, as Altman noted, 'designing a really good governance structure, especially for such an impactful technology, is not a one-week question. It's going to take a significant amount of time for people to think through this, to debate, to get outside perspectives, and for pressure testing.'
The exact nature of the internal changes at OpenAI remains uncertain. In a post on OpenAI’s website, Altman expressed his love and respect for Ilya Sutskever, stating, 'I think he's a guiding light of the field and a gem of a human being.' However, he also clarified that Sutskever would not have a seat on the new board, and discussions are ongoing regarding his continued involvement with OpenAI.
I believe this marks the end of the current drama at OpenAI, at least for the foreseeable future. As Altman mentioned in his interview with The Verge, the board will conduct an independent review, but it will take time for the results to be published. Then it will take even more time for a new governing structure to emerge. But at least for now, the situation has stabilised and people who work at OpenAI can focus on their primary mission: “to create safe AGI that benefits all of humanity”.
If you enjoy this post, please click the ❤️ button or share it.
I warmly welcome all new subscribers to the newsletter this week. I’m happy to have you here and I hope you’ll enjoy my work. A heartfelt thank you goes to all of you who joined as paid subscribers this week. Thank you for your support!
The best way to support the Humanity Redefined newsletter is by becoming a paid subscriber.
If you enjoy and find value in my writing, please hit the like button and share your thoughts in the comments. Additionally, please consider sharing this newsletter with others who might also find it valuable.
For those who prefer to make a one-off donation, you can 'buy me a coffee' via Ko-fi. Every coffee bought is a generous support towards the work put into this newsletter.
Your support, in any form, is deeply appreciated and goes a long way in keeping this newsletter alive and thriving.
🦾 More than a human
The XPRIZE Foundation is launching a new competition aimed at improving healthspan. The winning team of the 7-year-long, $101 million XPRIZE Healthspan competition must demonstrate that their therapeutic treatment restores muscle, cognitive, and immune function, by a minimum of 10 years, with a goal of 20 years, in individuals aged 65-80 years. The therapeutic treatment must show its effects in 1 year or less in a clinical trial. The competition is open to everyone, from biotechnology companies, academic investigators, research networks, and institutes, from within and beyond the field of ageing and geroscience. Teams can originate from any nation or combination of nationalities and can take the form of a for-profit corporation, a non-profit, or an academic institution. However, no government entities are allowed to register.
🧠 Artificial Intelligence
Millions of new materials discovered with deep learning
DeepMind is continuing its streak of causing breakthroughs in science with AI. After solving the protein folding problem with AlphaFold, and recently releasing a best-in-class weather-forecasting AI, the next target was materials science. With Graph Networks for Materials Exploration (GNoME), researchers at DeepMind were able to find 2.2 million new crystals – equivalent to nearly 800 years’ worth of knowledge. Of all these new crystals, 380,000 are the most stable, making them promising candidates for experimental synthesis. Now, it's up to materials science researchers to determine the viability and potential applications of these crystals dreamed up by an AI.
In this video, Andrej Karpathy, a well-respected AI researcher, gives an introductory talk to large language models - what they are, how they work and what improvements can we expect in the near future as well as how these models can be hacked. It is an excellent introduction to large language models and gives a better understanding of what these models are, what are their limitations and how those limitations can be overcome.
US, UK and a dozen more countries unveil pact to make AI ‘secure by design’
The United States, the United Kingdom, and 16 other countries recently released a 20-page agreement outlining guidelines for creating AI systems that are "secure by design" to prevent misuse by rogue actors. This non-binding agreement, emphasizing safety in AI development and deployment, includes general recommendations like monitoring for abuse, data protection, and software supplier vetting. The agreement, part of global efforts to guide AI development, doesn't address specific AI uses or data-sourcing methods. It suggests security measures like thorough testing before releasing AI models.
Amazon Q, an AI-powered chatbot for businesses
Amazon is launching an AI-powered chatbot for Amazon Web Services (AWS) customers called Q, which is currently in a public preview. Q can answer questions such as “How do I build a web application using AWS?” This chatbot is part of several new products and services coming to AWS, unveiled at the AWS re:Invent event. These include a text-to-image generator named Titan and new chips designed for AI training and inferencing. For more detailed information on the announcements, I recommend checking out this list from TechCrunch.
Sarah Silverman Hits Stumbling Block in AI Copyright Infringement Lawsuit Against Meta
A federal judge dismissed most of Sarah Silverman's lawsuit against Meta, challenging the use of copyrighted books for training its AI model. Judge Vince Chhabria rejected the argument that Meta's AI, or its outputs, constitutes derivative works that infringe on copyright, due to lack of evidence that they transform or adapt the original books. This decision mirrors a similar ruling in a case against AI art generators, emphasizing the need for substantial similarity between AI outputs and copyrighted works to prove infringement. While the judge dismissed several claims, he allowed for the possibility of repleading some. The ruling adds to the ongoing debate over the use of copyrighted material in AI development and its legal implications. Silverman has also joined a class action against OpenAI with similar claims.
$10 Million Artificial Intelligence Mathematical Olympiad Prize
XTX Markets is launching the Artificial Intelligence Mathematical Olympiad Prize (AI-MO Prize) with a prize pool of $10 million. The grand prize of $5 million will be awarded to the first publicly shared AI model that competes in an AI-MO-approved competition and achieves a performance level equivalent to receiving a gold medal in the International Mathematical Olympiad. The first AI-MO competitions are set to open for participants in early 2024.
If you're enjoying the insights and perspectives shared in the Humanity Redefined newsletter, why not spread the word?
Pentagon unveils ‘Replicator’ drone program to compete with China
Inspired by the effectiveness of drones in the Ukraine conflict, the Pentagon announced the "Replicator" program to deploy thousands of small, smart, and inexpensive autonomous systems in the next two years to compete with China’s similar drone programs. The program, primarily a reorganization of existing funds, is expected to cost hundreds of millions of dollars.
I didn’t expect to see boxing robots from Cyberpunk 2077 becoming a reality this quickly. The robot, developed by IHMC and Boardwalk Robotic, translates the movements of a human operator wearing a VR headset into its own movements. To make things more fun, the team has chosen boxing to showcase what this approach to controlling a robot can do.
Manchester researchers design and fly world’s largest quadcopter drone
Engineers at The University of Manchester have built and flown the world’s largest quadcopter drone. Made from foamboard, the drone measures 6.4m (about 21 ft) corner to corner and weighs 24.5kg (approximately 54 lbs), just 0.5kg under the weight limit set by the Civil Aviation Authority. This project is the follow-up to their previous endeavour, the world’s largest foamboard aircraft.
Researchers from ETH Zurich have created a new technique for 3D printing a robotic hand with bones, ligaments, and tendons in a single print. The new technique, which combines different types of soft, elastic, and rigid materials, offers advantages over traditional metal robots, particularly in safety and handling delicate objects. As someone who does semi-professional 3D printing, I am quite impressed by the results achieved by the team.
Italy bans cultivated meat products
Italy has become the first country to ban the production, distribution, and import of cultivated meat. The decision was driven by concerns for health, Italian production systems, job protection, and culinary heritage. This move contrasts with developments in other countries. Singapore has been selling cultivated meat since 2020, and the US approved the sale of cultivated meat to customers earlier this year. In Europe, the Netherlands has initiated pre-approved tasting sessions for cultivated meat in a controlled environment. Meanwhile, Romania is considering a similar ban to Italy’s, and in the US, Florida is proposing legislation to outlaw cultivated meat to protect local industries.
A Life-Extension Drug for Big Dogs Is Getting Closer to Reality
Instead of focusing on human longevity, a biotech startup Loyal aims to increase the life expectancy of large and giant dog breeds. The company has set a target to bring its life-extending drug to market by 2026. However, before achieving this, they must prove the drug's safety and manufacturing reliability to the U.S. Food and Drug Administration (FDA). Loyal has planned a comprehensive study for 2024 and 2025, which aims to enrol about 1,000 older dogs from large and giant breeds. This study is crucial for determining the drug's effectiveness in prolonging the dogs' lifespans.
PhD student bioengineers potato plant to detect gamma radiation
A PhD student at the University of Tennessee has developed a potato plant that can detect gamma radiation. Whenever the plant detects increased levels of radiation, it starts to glow green. This fluorescent glow can then be detected with cheap cameras. This innovation offers a simple, effective way to monitor radiation levels without complex equipment and provides an accessible, low-cost radiation detection method.
Thanks for reading. If you enjoyed this post, please click the ❤️ button or share it.
Humanity Redefined sheds light on the bleeding edge of technology and how advancements in AI, robotics, and biotech can usher in abundance, expand humanity's horizons, and redefine what it means to be human.
Humanity Redefined is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
A big thank you to my paid subscribers, to my Patrons: whmr, Florian, dux, Eric, Preppikoma and Andrew, and to everyone who supports my work on Ko-Fi. Thank you for the support!