We have achieved comprehensive artificial intelligence

A few days have passed since the full release of OpenAI’s o1 model, and now the company’s Iranian employee claims that we have reached Artificial General Intelligence (AGI).

“In my opinion, we’ve achieved AGI, and with o1 it’s even more obvious,” OpenAI employee Vahid Kazemi said in a post on X. Of course, Vahid Kazemi makes a note in the definition of AGI; According to him, if AI is “better than any human in every task”, in fact, the o1 model “does better than most humans in most tasks”.

In fact, Vahid Kazemi is not saying that the artificial intelligence of this company is more effective and skilled than an expert and skilled person in a specific task, in fact, he is saying that o1 can do many different things, even if the final result is not correct; However, no human can compete with the sheer breadth of new AI skills.

Opinions of an Iranian OpenAI employee about artificial intelligence

Vahid Kazemi has a PhD in machine learning and is a member of the OpenAI technical staff, and in his recent post he also discussed the nature of large language models, or LLMs, and whether or not they simply “follow a guideline.” He says:

“Some say that LLMs only know how to follow a guideline. “No one can really explain what a trillion parameter deep neural network can learn.”

He compares the instructions of artificial intelligence to the instructions of scientists; Kazemi says the entire scientific method can also be summed up in one guideline: observe, hypothesize, and verify. Good scientists can generate better hypotheses based on their intuition, but of course, this intuition itself is built with a lot of trial and error. In fact, according to Vahid Kazemi, although artificial intelligence may make mistakes now, it will eventually look like an expert scientist who puts forward new hypotheses.

OpenAI CEO Sam Altman also talked about AGI a few days ago. “My guess is that we’ll get to AGI sooner than most people in the world think,” he said. Of course, Sam Altman tried to reduce the expectations of users from AGI in his speech.