'We haven't seen a breakthrough': Salesforce AI chief unimpressed by OpenAI's latest model
OpenAI has called its latest LLM a "significant leap in intelligence". Salesforce's chief of AI research Silvio Savarese told Capital Brief he isn't so sure.
The chief of Salesforce’s global AI research thinks OpenAI’s latest foundation model represents only a modest improvement on the company’s previous efforts.
OpenAI described GPT-5 as a “significant leap in intelligence” when it launched the model in August, but Salesforce EVP and chief scientist of AI research Silvio Savarese told Capital Brief the improvements so far are more modest.
“We haven’t seen any breakthroughs there,” Savarese said. “There is a bigger context window and we saw some improvements with respect to multi-step reasoning… in terms of long-term planning and complex reasoning, that’s where things are more marginal.”
A model’s “context window” refers to the amount of text it can process, while “long-term planning” means its ability to understand how a problem or scenario changes over time.
