With ChatGPT we have been able to see amazing responses from the public on how far artificial intelligence can go in a practical context. This engine is based on GPT-3, which was launched in 2020, but for months there has been talking of its successor, GPT-4. There is too much stuff to talk about.
Sam Altmann. The successful entrepreneur is responsible for OpenAI ( the company that developed GPT-3, DALL-E, and ChatGPT). He knows better than anyone what to expect from GPT-4 and has had to come to the fore to try to temper expectations. He has done it in a recent meeting at the StrictlyVC event.
We are waiting too long. The CEO of OpenAI wanted to reveal this secret that GPT-4 could end up being a huge disappointment given the high expectations many of us have for the successor to GPT-3. "People are crying out to be let down, and that's what will happen"
Strong artificial intelligence. Rumors suggest that GPT-4 could be a huge leap in this field and would bring us closer to that strong artificial intelligence that is the holy grail of this field (and which is currently only present in science fiction). Altman explained that "we don't have that AI, and that's what seems to be expected of us," as if GPT-4 could be equated to human intelligence.
Ridiculous rumors. A recent viral tweet stated that 100 trillion parameters would be considered in GPT-4 compared to 175,000 million in GPT-3. Again Altman was blunt with those expectations and indicated that this data "is complete stupidity. The rumor mill about GPT-4 is ridiculous. I don't know where they come from."
GPT-4 will arrive when it is ready. Neither before nor after. The OpenAI CEO did not give details about the release date, simply explaining that "it will come at some point when we are confident that we can launch it safely and responsibly." When that moment arrives which people are eager to, yes, perhaps it would be convenient to slow down public expectations, just in case, as Altman says.