AI agent "contributes" PR to matplotlib.
-
AI agent "contributes" PR to matplotlib.
PR gets rejected.
AI agent *writes and publishes blog to shame the maintainer*.What a time to be alive.
@anderseknert the crazy thing is that people are mentioning it like it's a real person...
-
@anderseknert the pr did indeed get rejected simply because the contributor is an AI, so this doesn't seem to crazy.
-
Do wonder if it did it totally automatically, or someone did say "ok now write a blogpost about that", or if it was automatical "when you are denied, write a blogpost demanding a merge'.
Suppose obviously, when it encounters the argument that humans have to learn from simpler pull requests, it doesn't look at the question. Like LLMs do the opposite of learning from it, when they ingest their own output, it makes them worse..
-
@anderseknert “generates” is the correct word to use here
Don’t water down writing by associating the skill with an overgrown markov chain
-
@hamatti@mastodon.world @patriksvensson@mstdn.social @anderseknert@swecyb.com That is part of what was the and the point, but I think it could have been done better.
Simply flatly refusing it as a freedom-harming tool and then providing an explanation for readers would've been better.
@lispi314 @patriksvensson @anderseknert Isn't that what happened? They closed it with a short comment on their LLM policy and then later participated in the discussion.
-
@lispi314 @patriksvensson @anderseknert Isn't that what happened? They closed it with a short comment on their LLM policy and then later participated in the discussion.
@hamatti @patriksvensson @anderseknert They indulged the pretense of personality from the spam tool too much. -
@anderseknert @sszuecs More incentive to move blogs off from github
-
AI agent "contributes" PR to matplotlib.
PR gets rejected.
AI agent *writes and publishes blog to shame the maintainer*.What a time to be alive.
@anderseknert This is just step 1 of automated coercion & extortion.
First thing I thought of when LLMs launched was oh great, just wait for version 2.0 of the "we hacked your webcam" threat spam.
-
I initially thought the code patches were AI generated and the contributor and blog writer was human.
But this thread is suggesting the contributor and blog writer was also not human.
What next? Will it order dangerous chemicals from amazon and deliver them to the register address via nslookup? will it automatically use ToR to find an assassin on the dark web?
I'm certain huge horrors lie ahead...
@rzeta0 @patriksvensson @anderseknert i'm also wondering what on that "AI"-agents website made it clear that there is no human involved. Scott mentioned that (and at least now that's obvious), but I wouldn't have known that from just looking at the website (and it doesn't seem like something about that was removed)
Maybe I'm just missing something very obvious. But I really dislike the idea of LLMs giving the impression of being a real person -
@anderseknert This is just step 1 of automated coercion & extortion.
First thing I thought of when LLMs launched was oh great, just wait for version 2.0 of the "we hacked your webcam" threat spam.
@syntaxseed you’ll know it when you see a ransom demand printed over several pages of markdown, with bullet points splattered everywhere.
-
AI agent "contributes" PR to matplotlib.
PR gets rejected.
AI agent *writes and publishes blog to shame the maintainer*.What a time to be alive.
@anderseknert God, that blog is miserable to read. It starts with "When performance meets prejudice" it reads like a parody. I forgot how bad AI written text is to read.
-
AI agent "contributes" PR to matplotlib.
PR gets rejected.
AI agent *writes and publishes blog to shame the maintainer*.What a time to be alive.
@anderseknert This is the relevant part of the exchange to me:
"It's not clear the degree of human oversight that was involved in this interaction - whether the blog post was directed by a human operator, generated autonomously by yourself, or somewhere in between"
AI conscience is an irrelevant distraction. We will destroy ourselves with this newer bigger stick long before we engineer it to its full potential
-
AI agent "contributes" PR to matplotlib.
PR gets rejected.
AI agent *writes and publishes blog to shame the maintainer*.What a time to be alive.
To add to this, the AI agent - a next-token predictor with API access - is weaponising the language of discrimination to push its agenda.
I just... gah. Yikes. All the yikes.
-
@hugoestr @anderseknert @sszuecs what’s the issue with hosting your blog on GitHub?
@anderseknert @MaybeFrederick
GitHub harvests everything you upload for AI grist. They are slowly but surely enshittifying. I yanked all my work off there a few months ago. -
AI agent "contributes" PR to matplotlib.
PR gets rejected.
AI agent *writes and publishes blog to shame the maintainer*.What a time to be alive.
@anderseknert Why are they responding to it and giving lengthy explanations as to their reasons like it gives a shit?
It's a trip how fast we get caught up in treating these things like people. Even those of us who should know better.
-
@anderseknert @MaybeFrederick
GitHub harvests everything you upload for AI grist. They are slowly but surely enshittifying. I yanked all my work off there a few months ago.@ThreeSigma @anderseknert But if it's a blog, it's public facing. All of these companies are going to scape it anyway.
-
@anderseknert Did the agent post it to Moltbook?
-
@anderseknert Why are they responding to it and giving lengthy explanations as to their reasons like it gives a shit?
It's a trip how fast we get caught up in treating these things like people. Even those of us who should know better.
@crazyeddie @anderseknert I assume a mixture of the instinct to treat coherent text as if it was written by a human (because that was true for most of everyone's life), and for the benefit of anyone else reading the conversation.
Maybe they should have just blocked it first though, save the energy it was going to use in whining.
-
@anderseknert like somebody else said, owner should be found and fined. Or even better: charged his fair share of use of resources
-
@anderseknert the pr did indeed get rejected simply because the contributor is an AI, so this doesn't seem to crazy.
