- OpenAI may have fired Atlman because of its secret under-development project, suggests a new report.
- The company had reportedly discovered a new breakthrough AGI, capable of thinking like humans.
- Some staff feared that this new AGI would destroy humanity.
OpenAI has been the talk of Silicon Valley ever since its founder Sam Altman got fired last week. While Altman has already returned as CEO after Microsoft’s intervention in the matter, it is still unclear why he was fired.
As per OpenAI’s official statement, Altman was fired due to his lack of communication with the company. However, a new report from an anonymous OpenAI employee suggests that the company was close to developing a new artificial general intelligence (AGI) model codenamed Q* that could prove dangerous for humanity.
Altman’s aggressive approach to releasing new features could have caused OpenAI to deploy Q* quickly, and some employees saw it as a dangerous threat to humans. It is possible that this could be the real reason for Altman’s firing.
What is Q*?
Reuters says that an anonymous OpenAI employee got in touch with the news outlet, explaining the real situation at the company. The source says that Q*(pronounced as Q-Star) is an upcoming breakthrough in OpenAI’s research for AGI (artificial general intelligence).
While AI is powerful enough to learn through data provided to the model, AGI takes the concept several notches above. AGI is not just able to solve complex problems, but it can also generalise information and learn from its mistakes, just like a human being.
The source says that OpenAI was close to developing its AGI named Q*. It appeared to have the potential to surpass humans in economically valuable tasks. Q* was being considered as a breakthrough in OpenAI’s advancements.
At the time of testing, Q* was able to solve mathematical problems on its own, without any previous training. However, it was only solving math questions at a high school level. It was enough to impress researchers, while it worried others about its potential applications.
As word spread about Q* among more OpenAI staff, some members notified the board about the discovery of this new AGI. But rather than praising its discovery, several staff were afraid of it and reported it as a huge threat to humanity.
OpenAI Board Feared Altman Over Q* Public Release
Altman is known as a risk-taker and has deployed a lot of features into production without worrying much about consequences. He himself admitted to pushing the veil of ignorance multiple times in his stint at OpenAI so far.
The OpenAI board started to get concerned over Altman’s approach and feared that Q* would be released without exploring the negative outcomes. The source says that researchers spoke to the board that AGI could reach a point where it could decide that the destruction of humanity was in its interest.
So Q* is all the rage now eh?
I wanted to know how LLMs could benefit from Q learning + A* search so I asked chatgpt to explain it to me.
Imagine you're playing a video game where you need to find treasure in a maze. In this game, Q-learning is like having a super…
— Dylan@SOG (@sog_on_bird_app) November 23, 2023
Within less than a year of ChatGPT’s launch, the service has shown its ill effects. Students have used ChatGPT as a shortcut for their assignments, while professional employees were also found to be using it at their jobs.
OpenAI’s Q* AGI could worsen the situation if it was released without thinking about all possible outcomes. But the problem is, it’s almost impossible to rule out all drawbacks in the case of AGI, which almost thinks like a human.
This may have led the OpenAI board to reconsider Altman’s leadership, and eventually fire him from the company. However, OpenAI has not yet acknowledged the presence of Q* or any other AGI model.
As of now, Sam Altman has returned to OpenAI as the CEO. Before getting fired, Altman had mentioned an upcoming new technology during one of his conferences. Now that he is back, it is possible that we may officially hear about Q* directly from the company.
However, considering that Microsoft is now involved in OpenAI’s management closer than ever, there is also a possibility that the company may become more mature in its upcoming public releases. Until then, Q* and its supposed capabilities will continue to remain an unsolved mystery.