OpenAI's battle drama ended with the return of Sam Altman, but inside OpenAI, the truth surrounding Altman's dismissal also gradually surfaced.。

OpenAI held a secret meeting to discuss whether Altman would stay or not.

On November 22, sources noted that four days before OpenAI's board decided to exile Altman, several of the company's senior researchers had written to the board warning of a powerful AI discovery.。They said the discovery could threaten humanity.。

So the board organized a secret conversation in which the researchers expressed concerns about a potential breakthrough in artificial intelligence.。They argue that not only did OpenAI CEO Altman deliberately hide the speed of the project's progress in this study, but Altman's other actions also seriously violated OpenAI's original intent, including prematurely commercializing new AI technologies and being stubborn on major strategic issues for the company.。

After much speculation, it was at this secret meeting that the board made up its mind to fire Altman because the above allegations corresponded to what the board said when it fired Altman that it had "not always been candid in communicating with the board."。

Q * has been superior to humans in certain aspects

According to OpenAI insiders, the technical code name of the artificial intelligence that directly led to Altman's departure is Q *。

There are currently three explanations for what Q * is.。The first explanation is that Q * may be synonymous with Q-Learning; the second explanation is that Q * is the code name of a new model created by OpenAI with the help of Q-Learning algorithms; and the third explanation is that Q * is simply a project name within OpenAI.。

Insiders also say that the biggest feature of Q * is that it outperforms humans in most tasks that require consideration of economic significance。For example, Q * is likely to be adept at solving complex multi-stage problems, i.e., dynamic programming equations。Usually in this equation, the person (or computer) running the algorithm can enter an objective function, such as "shortest travel time, lowest cost, greatest profit, greatest utility," etc.。The algorithm will then decide what best action to take to achieve the desired result。

Q *, with its massive computing resources at OpenAI's disposal, has been able to solve certain mathematical problems with built-in models and has shown to be superior to human autonomous systems in certain areas.。Before Altman's dismissal, OpenAI conducted an internal demonstration of Q *, showing that Q * was already able to solve math problems at the elementary level, the source said.。Even at this stage, it seems that there is nothing special about Q * being able to solve elementary school problems, but the researchers are already very satisfied with the progress Q * has made at this stage and are very optimistic about its future success。

Q * or will enter artificial super intelligence in the future

Q * may mark the next level of OpenAI research。

In general, in the field of generative artificial intelligence, there may be multiple answers to the same question - large models will write and translate languages for users through a variety of statistics and predictions.。However, in the mathematical reasoning mode of Q *, there will only be one answer to a question, which greatly reduces the probability of AI interaction divergence。The fact that Q * "holds the only answer" suggests that AI will be more likely to reason like humans, and that the output of reasoning will be more rigorous.。

At present, there are already media discussions on whether Q * can enter the category of ASI (artificial super intelligence) in the future.。The concept of ASI implies that artificial intelligence already has reasoning capabilities close to human intelligence, able to understand, learn and apply its own knowledge like humans, and can even replace humans in various situations.。And, with ASI's capabilities fully realized, the AGI model can theoretically perform any intellectual task a human can perform, a potential threat that many AI experts have warned about.。

Speaking at a public event just before his dismissal, Altman said: "We've had four breakthroughs in the history of OpenAI, most recently in the last few weeks.。"

At present, the so-called "fourth breakthrough" is likely to refer to the emergence of Q *。