Sam Altman looks ahead to the future of ChatGPT. According to Altman, the GPT-4 model is not yet very smart, but that will quickly improve with the arrival of GPT-5.
To a novice in artificial intelligence, GPT-4 looks impressive, but OpenAI CEO Sam Altman sees it completely differently. During a lecture at the prestigious Stanford University in California, Altman made time for a Q&A about the future of AI and what’s in store for OpenAI. Altman was sharp and did not shy away from big statements.
For example, he openly expressed reservations about the intelligence of GPT-4. “GPT-4 is the dumbest model any of you will ever have to use again. It is important to deliver early and we believe in iterative implementation. If we build AGI in a basement and the world follows along blindly, I don’t think we’re good neighbors,” Altman reportedly said, according to Tom’s Guide.
Smarter GPT-5
This will improve soon. GPT-5 is in the waiting room and the new generation of the LLM may be made public in the summer. Altman can confirm ‘with a high degree of scientific certainty’ that GPT-5 will be much smarter than GPT-4, just as he already knows that a possible GPT-6 will be much better.
Altman and OpenAI dream of developing general artificial intelligence that is as capable as the human brain in all tasks and with GPT-5 they hope to get one step closer to that. That costs the company tons of money: OpenAI spends about $700,000 a day to keep ChatGPT running, and the total bill came to $520 million last year. Altman himself says that he is not interested in money, although it almost cost him his job last year.
When OpenAI will release GPT-5 to the world remains a matter of speculation. According to https://twitter.com/apples_jimmy/status/1785436609095328067 the company may first announce an AI-based search engine this week, not coincidentally just before Google holds its I/O conference. But these are also nothing more than rumors that are neither confirmed nor denied by OpenAI.
