Input Stories

Artificial intelligence models that take longer to consider requests made of them and the answers they give could be the key to cutting down on erroneous responses (The Wall Street Journal, subscription).

What’s going on: “Long-thinking AI models are designed to take more time to ‘think over’ the results they generate for us. They will be intelligent enough to give us updates on their progress and ask us for feedback along the way.”

• That could mean a few extra minutes—or months, as tech executive Jensen Huang said recently.
• “In many cases, as you know, we’re now working on artificial intelligence applications that run for 100 days,” Huang said at a trade show in Taipei last summer.

Systems 1 and 2: “The idea of long thinking builds on a dichotomy in human thought that the late Daniel Kahneman referred to as System 1 and System 2,” with the former being automatic and quick and the latter “allocate[ing] attention to the effortful mental activities that demand it, including complex computations.”

• Current generative AI is like System 1—but long thinking is an effort to move it into System 2 territory.

What’s to come: “The reasoning capability of the new models is still in the early stages, but is on track to make significant advancements next year,” according to an executive at OpenAI, the company behind ChatGPT.

• The increasing number of applications being built on System 2 could help repay some of the huge investment being put into AI.