Project Q*: self-learning AGI

In mid-November there was significant turmoil in the AI world. Sam Altman, CEO of OpenAI, was fired without a clear explanation by the board. Things have settled down since, but what exactly happened? And what does Microsoft have to do with it? What did we learn about Project Q*? Let me walk you through this strange episode.

OpenAI, founded in 2015 by Elon Musk, Sam Altman, and others, is a leading AI research organisation. Its goal is to develop ethical, human-friendly AI, with an emphasis on safety and societal benefit. OpenAI is known for its advanced machine learning models, including the GPT series that powers ChatGPT. OpenAI and Microsoft have a strategic partnership, with Microsoft investing in OpenAI (a 49% stake worth 13 billion US dollars) and providing cloud computing support through Azure. The collaboration lets OpenAI develop and train large-scale AI models like GPT, while Microsoft gets access to advanced AI for its own products and services.

On Friday 17 November, Sam Altman was suddenly fired by OpenAI's board. The board refused to give a clear explanation, only stating: "the board no longer has confidence in his ability to continue leading OpenAI". Almost immediately after the dismissal, it became known that Sam Altman would be welcome at Microsoft to lead a new advanced AI team there.

By Monday 20 November, all OpenAI staff were aware of the situation. They were surprised by the firing and very much disagreed with the board's decision, because Sam was widely liked inside the organisation. They pushed back and demanded the following: Sam Altman had to be reinstated as CEO of OpenAI and every member of the board had to resign. If the board did not agree, the staff would walk. The remarkable thing was that Microsoft had already pledged that any employee who resigned would have a good role on Sam's new research team at Microsoft. With nothing left to lose, 505 of OpenAI's 700 employees had signed the petition by Monday 20 November, meaning 72% of staff were prepared to resign if Sam Altman was not reinstated and the board did not step down.

On Wednesday 22 November, Sam Altman was reinstated as CEO of OpenAI and the board members resigned. You might think things returned to normal at that point. But during the chaos a major discovery surfaced. On Thursday 23 November it became known that Sam had been working on a secret project inside OpenAI. He had had a major breakthrough and apparently had not informed the board sufficiently, which was likely the cause of his firing. The project was Project Q* (pronounced 'Project Q-Star'), aimed at developing AGI.

Artificial General Intelligence (AGI) is a form of AI that matches or exceeds human cognitive ability, with the capacity to understand and tackle a wide range of tasks and problem solving. Unlike specialised AI, AGI can adapt to new situations and learn independently, the way humans do. Sam has reportedly used Project Q* to solve elementary school maths. That may sound simple, but it shows that Project Q* can solve maths problems it has never seen before: it learns from its mistakes and improves itself, like a human. Given how fast AI is advancing, Project Q* will quickly move on to higher mathematics. OpenAI staff and key researchers in the AI field have called Project Q* 'a threat to humanity'.

The expectation is that Project Q* will understand high-level mathematics by early next year and, by the end of next year, will be able to answer existential questions like 'why is there life?' How far this development can really go in answering such questions, and when, is still very much a matter of speculation. What is certain is that we stand at the beginning of an enormous evolution, the biggest in human history. The important thing is to stay informed about the latest developments and to apply rigorous standards to security and privacy across every new technology.