So, the thing is, generative AI makes stuff up.
I’m Jed Hartman, and I’ve been posting examples of generative AI making stuff up since 2022, mostly on Facebook. In late 2025, I decided to turn the posts into a site of their own.
So far, I haven’t received any questions, so I can’t provide an FAQ as such. But here are some questions that I expect to get, with answers:
About the site name
Isn’t the word lie kinda anthropomorphizing?
Yes. But it’s shorter and less clunky than hallucination (which is also anthropomorphizing). And it has suitably negative connotations. And I haven’t yet heard a term that I really like, other than makes stuff up.
Why not bullshit?
I don’t much like that word, and I don’t make the distinction between lying and bullshitting that some people make.
Why gen AI in the site name?
Yeah, I worry that that could be misinterpreted as meaning General Artificial Intelligence, which is a whole other issue. But gen is shorter and less clunky than generative. I’m going for a short catchy site name, and someone on Facebook suggested #GenAILies as a hashtag, and even though that phrase has various drawbacks, I feel like it’s short enough and catchy enough for my purposes.
About AI
What about all the other issues with generative AI?
I have lots of issues with generative AI. But at least for the moment, this site is specifically about cases where an AI system makes something up and presents it as truth. (Though my boundaries for that topic are a little fuzzy; sometimes I also include things like AI doing catastrophically bad stuff, especially if it then lies about what it did or why it did it.)
There are other kinds of AI, you know!
Yeah, this site is specifically about generative AI (and mostly about Large Language Models as such). Other kinds of AI are more likely to be beneficial and less likely to give blatantly false answers. (Though there can also be plenty of pitfalls in training and using non-LLM AI systems.)
About the posts
What’s with the Facebook links?
The first several dozen posts on this site originated as Facebook posts. In many cases, the original FB post has comments that provide further elucidation, context, jokes, etc. So for each post here, I’m including a link to FB. The FB posts are all public; you can view them even if you don’t have a FB account.
What’s with the Twitter/X links?
I don’t want to send any traffic to X, so I’ve replaced Twitter links with the equivalent links to xcancel.com.
What’s with the broken links?
In cases where I’ve noticed a broken link, I’ve replaced the original link with a link to the Wayback Machine. But there are probably places where I haven’t noticed a broken link. Alas, links often disappear over time.
About the design
Why is the site design so minimalist?
I just used WordPress’s default 2025 theme. I’m not a graphic designer and I wanted to get the site up and running quickly. I welcome recommendations for improvements, as long as they won’t cost me a lot of time or work.
Contact
Can I send you more examples?
I’m a little worried about getting overwhelmed with examples—there are so many! But for now, sure, you can send them to me in email at logos-gal@kith.org. But remember, this site is specifically for examples of generative AI providing false information and claiming it’s true. Also, please send links and/or text whenever possible; I dislike screen snaps of text, and would rather not include them on the site if I can avoid it.