学进去

For Kurt Gray, a social psychologist at the University of North Carolina at Chapel Hill, conducting experiments comes with certain problems. Before starting any study, his lab must get ethical(伦理的)approval from an institutional review board, which can take weeks or months. Then his team has to hire online participants—easier than bringing people into the lab, but Gray says the online subjects are often lazy. Then the researchers spend hours cleaning the data. But earlier this year, Gray accidentally saw an alternative way to do things.

He was working with computer scientists at the Allen Institute for Artificial Intelligence to see whether they could develop an AI system that made moral judgments like humans. But first they figured they’d see if a system from the startup Open AI could already do the job. The team asked GPT-3. 5, which produces human-like text, to judge the ethics of 464 scenarios(情境), previously evaluated by human subjects. It turned out that the system’s answers were nearly the same as human responses.

“This is crazy,” Gray says. “If you can just ask GPT to make these judgments, why don’t you just ask GPT instead of asking people?” The results were published this month in Trends in Cognitive Science.

Now, researchers are considering AI’s ability to act as human subjects in fields such as psychology, political science, economics, and market research. No one is yet suggesting that chatbots can completely replace humans in behavioral studies. But they may act as convenient stand-ins(替代者) in pilot studies and for designing experiments, saving time and money. Language models might also help with experiments that would be too impractical, or even dangerous to run with people. “It’s a really interesting time,” says Ayelet Israeli, a marketing professor at Harvard Business School who believes the models’ impact on behavioral research could amount to a “revolution”. “Some of these results are just astonishing.”