It's so frustrating that we have replicated the same errors of early web infolit by teaching people to look for "clues" that things are AI -- when we all knew those clues would slowly become non-indicative. 100% an unforced error, but people couldn't get enough of it. www.nytimes.com/interactive/...
The lack of durability of AI "clues" is the most predictable thing in the world. And so we've trained a generation of people on techniques that will now mislead them. And continue to do so! And no one seems to care.
Anyway, if you're teaching people to judge things by checking surface signals and surface signals are quickly evolving, please take a minute to think why that might be bad.
We're only at the start of this, and whatever you can do now won't be what you can do tomorrow. Read Gigerenzer on heuristics. Learn what a stable and determinative heuristic looks like. Or Origgi on signals.
