> Participants weren’t lazy. They were experienced professionals. But when the tool responded quickly, confidently, and clearly they stopped doing the hard part.

This seems contradictory to me. I suspect most experienced professionals start with the premise that the LLM is untrustworthy due to its nature. If they didn't research the tool and its limitations, that's lazy. At some point, they stopped believing in this limitation and offloaded more of their thinking to it. Why did they stop? I can't think of a single reason other than being lazy. I don't accept the premise that it's because the tool responded quickly, confidently, and clearly. It did that the first 100 times they used it when they were probably still skeptical.

Am I missing something?

The idea that everyone is either full lazy or not lazy is a bit reductionist. People change their behavior with the right (or wrong) stimulus.

Also, I won't remotely claim that it's the case here, but external pressures regularly push people into do the wrong thing. It doesn't mean anyone is blameless, but ignoring those pressures or the right (or wrong) stimuli makes it a lot harder to actually deal with situations like this.

> I suspect most experienced professionals start with the premise that the LLM is untrustworthy due to its nature.

Most people don't actually critically evaluate LLMs for what they are, and actually buy into the hype that it's a super-intelligence.

Could have performed accurately in their past usage, building trust. Sometimes it will also get something right that is downright shocking, far beyond what you hoped.

It's deceptively easy to trust the AI when it gives you mostly plausible answers.