CEO Sam Altman returns to OpenAI’s Board

 Sam Altman is a force of Nature. Very few people become more powerful by getting fired, but Altman managed just that by turning the tables on the OpenAI Board that fired him. He orchestrated a massive revolt among OpenAI employees, which forced the Board to reverse themselves and resign. Sam was not allowed back on the Board until an outside law firm investigated whether he had done anything naughty to cause the Board to give him the axe. The verdict is back, Sam was deemed (surprise! surprise!) clean as the proverbial whistle, and now Sam is back on the Board.

Even Sam’s closest friends are a little afraid of him. Paul Graham, legendary tech entrepreneur and founder of the fabled startup incubator Y Combinator, chose his protege Sam as his successor at that organization in 2014. Five years later, Graham fired Sam, an incident neither will talk about. Back in 2016, the New Yorker had published a profile of Sam at Y Combinator. The article’s title? Sam Altman’s Manifest Destiny. It was obvious even then.

Sam Altman returns to OpenAI’s Board

Read the 2016 New Yorker profile on Sam at Y Combinator

Elon pinky-swears that he will open-source his AI model, Grok

Elon sued OpenAI for being too focused on profits. Then OpenAI showed emails from Elon that proved he was a liar and a hypocrite. The New York Times published an article alleging that Elon’s charity mostly benefits himself. So now the world’s richest man-baby tries to grab back the moral high ground by promising to make his AI model, Grok, open source. Elon’s relationship to the truth is complicated at best, so don’t hold your breath while waiting for him to make good. And if and when he does, look for the very wide streak of self-interest that is sure to be included.

Elon pinky-swears that he will open-source Grok

Nvidia and HP join forces to produce AI laptops

Currently, AI runs mostly in the cloud, on massive servers provided by Amazon, Microsoft, or Google. AI chip-maker Nvidia and consumer laptop-maker HP want to change that. Each would like to see AI-on-a-laptop become a reality, so they could sell more hardware. Now they have announced an alliance to make that happen. They argue that running AI locally rather than in the cloud will increase speed and security, while reducing cost and energy. And revenue for Nvidia and HP? That would go way, way up, natch.

Nvidia and HP join forces to produce AI laptops

Hugging Face and former Tesla scientist team up on robots

Remi Cadene, an AI scientist formerly at Tesla working on self-driving vehicles and humanoid robots, announced on X/Twitter last week that he was joining Hugging Face, famed repository for open-source AI tools and models. Hugging Face has been primarily associated with open-source Large Language Models such as Llama 2.0 from Meta/Facebook, and Mistral from the French AI company of the same name. Now it appears that Hugging Face is preparing to move into AI for robotics, a space that has been heating up lately with news such as the recent $675 million raised by robotics startup FigureAI, valuing that company at $2.6 billion

Hugging Face and former Tesla scientist team up on robots

Florida middle schoolers arrested for deepfake nudes of classmates

Two middle schoolers from Pinecrest Cove Academy in Miami have been arrested for using AI to produce deepfake nude images of 12- and 13-year-old schoolmates. The teens are being charged with a third degree felony under a 2022 Florida law. Although the problem of deepfake nudes of classmates is a problem faced by any number of schools in the US, this appears to be the first instance in which the child perpetrators are being charged with a felony.

Florida middle schoolers arrested for deepfake nudes of classmates

Five of this year’s 45 Pulitzer Finalists are AI-powered

The Pulitzer Prize is journalism’s most prestigious award. This year, for the first time, 5 of the 45 Finalists disclosed that they used AI in the process of researching, reporting, or telling their submitted story. Entrants are allowed to use AI in crafting their stories, but are required to disclose it. The Pulitzer Board has allowed the use of AI because they felt that banning it would discourage newsrooms from engaging with innovative technologies. Pulitzer winners, along with the complete list of Finalists, will be announced May 8.

AI used by 5 of 45 Pulitzer Finalists

Korean researchers announce “neuromorphic” AI superchip

Researchers at the Korean Advanced Institute of Science and Technology (KAIST) have announced development a new “neuromorphic” superchip for running AI models. KAIST claims that the chip, whose architecture mimics in some ways the architecture of the brain, is 41 times smaller and uses 625 times less power than Nvidia’s workhorse A100 GPU chip. If these claims hold up (KAIST’s press office is sometimes a bit over-enthusiastic), chips based on this architecture might be able to run ChatGPT on mobile devices, making our smartphones waay smarter.

KAIST announces “neuromorphic” AI superchip

UK researchers use AI to decipher dark energy

Physicists now think that “normal” matter makes up only about 5% of the universe. Another 25% is made up of “dark matter,” which so far has only been detected by its gravitational effects. And a whopping 70% appears to derive from “dark energy,” the mysterious force causing the expansion of the universe. Recently, scientists at University College London have used AI to infer the workings of dark energy in the evolution of the visible universe, and report that this new AI system is 4 times more powerful than previous methods.

UK researchers use AI to decipher dark energy

Google develops AI to read prenatal ultrasounds for rural Africa

Maternal mortality in childbirth in Sub-Saharan Africa is among the highest in the world. Pregnant women in rural Africa may live far from a hospital, and the local health worker is likely to be untrained in reading the prenatal ultrasounds so important to assessing the health of the mother and the growing fetus. Google is partnering with Jacaranda Health, a Kenya-based nonprofit to develop an AI system that can help local health workers read prenatal ultrasounds as well a trained sonographer.

Google developing AI to read prenatal ultrasounds for rural Africa

China developing AI chatbot to assist neurosurgeons

Chinese developers are testing an AI chatbot to assist neurosurgeons with diagnosis and treatment planning for neurosurgical patients. Specialists in Hong Kong have collaborated with AI scientists and engineers in building a multimodal advisor based on the open-source Large Language Model Llama 2.0 developed by Meta/Facebook. Progress on the project has been hampered by US bans on export of advanced computer chips, but testing is rolling out at several centers this year.

China developing AI chatbot to assist neurosurgeons

MIT and Mass General develop AI to predict drug interactions

Researchers at MIT, Mass General Brigham, and Duke University have developed an AI model that helps identify the specific transport protein used by a given drug, in order to avoid cross-drug interactions with medications competing for the same transport protein for absorption into the circulation across the gut lining.

MIT and Mass General develop AI to predict drug interactions

South Korea startup launches $1800 AI doll companion for elders

A South Korean startup has developed an AI-powered companion doll designed to offer companionship and emotional support to the country’s burgeoning elderly population. The Socially Assistive Robot (SAR) has been independently studied by researchers, and improvements in depression, cognition, and medication adherence have been found. Local governments, faced with a rapidly-aging population, have been testing these robots as a cost-effective way to support seniors who have limited social support.

South Korea startup launches $1800 companion doll for seniors