Hey, survey researchers? Don't do this.
Headline in my inbox this morning: "23% of U.S. adults now use AI language models like ChatGPT – the tipping point."
Actual survey question (which I had to click a link and read 17 pages deep to find): "Have you ever used an AI language model or a chatbot."
Exactly right. I tried one out, and it was fine enough. But I don't need it now.
For all the hilariously-bad results that AI can deliver, the biggest issue is when people find it's a fun toy but not actually useful.
Honestly, the fact that, after 2 years of hype, the answer to "have you EVER..." is only 23% is a HUGE data point against the idea that AI is at a "tipping point" for going mainstream.
If you can't tell the difference btwn "have you ever tried <the giant tech fad of the past 2 years>" and "do you routinely use <giant tech fad>" then you should just give up and replace yourself with ChatGPT.
It'll hallucinate and be at-best-average at your job. But that'd still be an improvement.