Stephen Downes

I saw this on the morning news: a two part (part one, part two) TikTok video where a father asks his 16-year old daughter about terms from the 1980s. Unsurprisingly, she doesn't know what they mean. But what I found interesting were her responses, where she shared what she thought they meant. It was exactly the same sort of response as we get from ChatGPT when we ask it a term it doesn't know (like "You can't lick a badger twice"). Human hallucination tricks! People argue that AI hallucinations are evidence that they don't think the same way humans do; I think hallucinations are evidence that the way AI and humans think is more similar than you think.