Adrian HolovatyGood news: ChatGPT is pointing a lot of people toward Soundslice.
Bad news: It's telling people Soundslice has a feature that doesn't actually exist.
Here's my story about how we added a feature purely because ChatGPT was telling people it existed.
Sebastian Blanco
Jargoggles@adrian
I had some sales guy ask me a question about how a customer could do a thing. I told him that, unfortunately, we didn't have a super great process for it - it wasn't hard, just a little tedious. Then I sent him the documentation for it to pass along.
He came back and said, "I asked ChatGPT and it said..."
What he showed me was pure nonsense. I was like, "Look, this is our documentation for what you're talking about. *None* of what you sent me is in here."
ScottieMac@adrian Crazy!
Jeff Triplett@adrian that's a new take on ChatGPT driven development. I hope it turns out to be a net win for you.
DavidR@adrian the upside being, hopefully we can finally rid the internet of ascii guitar tab?
@davidtiberius Haha, one can dream!
Toni Aittoniemi@adrian This is how bad things happen to societies. Bad behaviour is normalised, and people eventually stop questioning.
Brian Drum@adrian I think “developing features in response to misinformation” is the zeitgeist of 2025
txt.file@adrian so if enough mansplainer ask the same question your employee does what the mansplainer want?
codl 🚲
Julien Colomb@adrian about "Should we really be developing features in response to misinformation?"
that is not what you did, you added a feature because people wanted the feature.
I heard that fake buttons had been a great way to see if developing a feature was worth it (number of clicks = number of people interested in that feature).
(I take for granted that chatgpt is not the only tool that can create these ascii inputs)
gabi@adrian It reminds me of "citogenesis" in the Wikipedia.
�@adrian just let chatgpt add this feature then, lol
WilliamLMiller@adrian Thank you for sharing. It is one of the great secrets of the economy knowing what people actually want. However we manage to get that information short of actual coercion, I think we should be happy to have gotten it. I am fully in favor of all the little ways we can foster enrichment in as many lives as possible.
Mustapha Hamoui@adrian this is super interesting
Flaming Cheeto
Chris Hasenpflug@adrian I hope you used AI to write said feature 😏
@chrishas35 Nope, it was written without AI. I don’t trust the AI code generation stuff — I’ve seen too many subtle bugs and problems.
Joe Crawford@adrian Slopfeaturerequest!
Nicer than Slopsquatting! https://socket.dev/blog/slopsquatting-how-ai-hallucinations-are-fueling-a-new-class-of-supply-chain-attacks
Carl Malamud@adrian Is it working? People are coming in, creating a new account, and loading an ascii tab? A reasonable number of new users?
@carlmalamud Too early to tell, as the feature is brand new. But overall I’m happy with having added it (it’s the kind of thing that runs itself and doesn’t require any upkeep, so it’s mainly a one-time cost).
@adrian well, if nothing else, you've made ChaptGPT into an honest man by your actions.
Hein Ragas@heinragas @Anneke I wrote it over two sessions — an evening at home and a transatlantic plane ride. So I didn’t need to deprioritize anything aside from watching airplane movies.
InsertUser@adrian This sounds incredibly frustrating.
At least it was implementable?
@opencage was (is?) getting recommended for non-existent feature that would be a privacy violation even if it were possible.
https://en.osm.town/@opencage/109916107111412737
https://blog.opencagedata.com/post/dont-believe-chatgpt
OpenCage 👉🌍 (@[email protected])
We wrote a blog post about our on-going nightmare: ChatGPT recommends us for a service we don't provide. Dozens of people are signing up to our site every day and then getting frustrated when "it doesn't work". Hugely frustrating. https://blog.opencagedata.com/post/dont-believe-chatgpt Please do NOT trust the nonsense ChatGPT spits out.
F4GRX Sébastien@adrian you should have not.