What you need to know
- OpenAI CEO Sam Altman says that ChatGPT is the dumbest it will ever be right now.
- He promises better performance and accuracy with the long-anticipated GPT-5 model.
- Altman had previously indicated that OpenAI’s GPT-4 model “kind of sucks.”
OpenAI CEO Sam Altman was at Stanford University earlier this week, where he had a lengthy discussion about his vision for the company and future advances in AI (via Tom’s Guide). As you may know, Altman has been championing the AGI superintelligence beat for a hot minute now.
It takes an arm and a leg to keep a supercharged AI-powered chatbot like ChatGPT on a daily basis. Last year, we learned that OpenAI spends up to $700,000 per day to keep its ChatGPT operations running. Interestingly, the cost implication to keep these operations afloat doesn’t seem to be a major concern for Altman. Instead, he’s more focused on providing users with capable tools that will make work easier and let them achieve incredible feats.
A ChatGPT that’s “really good, like materially better”
While speaking about ChatGPT, Altman boldly said, “GPT-4 is the dumbest model any of you will ever have to use again by a lot.” It isn’t the first time the CEO has made such a comment about the model. He admitted GPT-4 “kind of sucks” while rumors swirling the air indicating OpenAI is working on a new model, potentially GPT-5, which is expected to be “really good, like materially better.”
When asked about the cost of running each subsequent model that OpenAI unveils and how it capitalizes on it, Altman indicated:
“I think giving people really capable tools and letting them figure out how they’re going to use this to build the future is a super good thing to do, and is super valuable, and I am super willing to bet on the ingenuity of you all and everybody else in the world to figure out what to do about this. There is probably some more business-minded person than me at OpenAI somewhere that is worried about how much we’re spending, but I kind of don’t.”
Altman was also asked about spending over $520 million on ChatGPT last year and indicated GPT-4 is “mildly embarrassing at best.” He added the company is working towards improving the LLM’s performance and accuracy with new versions of the model.
GPT-4 is the dumbest model any of you will ever have to use again by a lot. It’s important to ship early and often and we believe in iterative deployment.
OpenAI CEO, Sam Altman
Altman concluded his remarks by indicating with “a high degree of scientific certainty that GPT-5 is going to be a lot smarter than GPT-4.” He further added that GPT-6 would, naturally, be better than GPT-5, indicating the nature of AI and software development in general.