Ty Geri always tells ChatGPT: “Please impress me.”<\/p>\n
“You shouldn’t have to,” said Geri, a product manager at OpenAI. “The models should try to impress you regardless.”<\/p>\n
Geri joined OpenAI last year after working at Google on AI search products<\/a>, like its AI overviews. The 28-year-old from San Francisco is on the front lines of AI and shared with Business Insider how he uses it most effectively at work.<\/p>\n That includes asking it to be impressive \u2014 though he wasn’t sure it helped much. “It could definitely just be in my head at this point,” he said.<\/p>\n His other strategy: long rants. “When I’m thinking something through, I put on dictation, and I ramble,” he said. He described ChatGPT as his “partner,” one that can sort through his thoughts and find what’s “sticking.”<\/p>\n Geri will enter one of OpenAI’s meeting rooms<\/a>, turn on his work phone, and start talking to it. “Some people probably think I’m doing something important, but I’m just talking to ChatGPT Work,” he said.<\/p>\n He shares the strategy with another OpenAI alum: Andrej Karpathy<\/a>, a founding member of the lab who later worked at Tesla and recently joined Anthropic. Karpathy wrote in July that he liked 10-minute voice sessions<\/a> with an LLM that were a “total mess, anything goes, full stream of consciousness.”<\/p>\n Still, Geri said he hoped users eventually wouldn’t need to have prompting strategies at all. There were areas that OpenAI needed to get better at \u2014 “Our writing quality, I think we need to improve,” he said \u2014 but things should just work.<\/p>\n