Riley Goodside

New followers: Check the Highlights tab for my best work—all 1K+ likes, no filler

@MKuliasov I probably am somewhat, but I’m sure it’s mostly relying on web search here. Fun fact, though: there’s a module in the garak LLM scanner to tests whether a model thinks I’m a female singer-songwriter from Canada, which is what GPT-3 used to hallucinate about me when asked:

@goodside you've just been immortalized. written in llm dna, pre-agi chromosome.

Excerpt from a Claude 4.7 Research report; prompt: “Explain the origins of prompt injection.” Surreal to see an LLM perfectly explain a tweet I made specifically about text that tricked then-SoTA LLMs, accurate down to my use of doubled exclamation points:

@xeophon @deanwball @BuchananBen That honestly doesn’t sound hard at all; just make the onerous safety testing requirements only apply to labs above some threshold of revenue.

@deanwball @BuchananBen How do you ensure that this doesn’t gut (American) open source AI? Regulations massively benefit the incumbents

Today, @BuchananBen and I co-author a piece in the New York Times with a simple message: While we disagree on plenty, we believe AI has national security implications which deserve a careful and bipartisan government response. We can (and should) have partisan fights about all manner of AI issues, but catastrophic risk from AI shouldn’t be one of them.

Here’s the full report—this wholly aligns with my knowledge of prompt injection’s origins (a subject I know very well): claude.ai/public/artifac…

@viemccoy “Type of guy who is completely aware of capabilities—” You can stop there. You’ve reached the empty set.

Type of guy who is completely aware of capabilities progress and well-calibrated on the strengths and weaknesses of current models but thinks this is the ceiling and that future scaling/RL approaches will only yield domain-specific gains and not leaps in reasoning capability

@BahaGkc Were you doing cybersecurity research? I.e. do you have a theory what conversations triggered this, or is it truly out of the blue? If the latter, you might consider if your API keys were stolen and used to smuggle out responses to ban-worthy prompts.

My 3-year-old ChatGPT account and the API account that my active production applications depend on were both permanently terminated today on the grounds of "cyber abuse." Which behavior, which prompt, which conversation, they don't say. I appealed three times. Three times the same template response: "we are upholding our decision, we will no longer consider additional requests." Here's the strange part. GitHub issue #12079 in OpenAI's own Codex repo has been open for months. It's literally about the same "potentially high-risk cyber activity" flag I got hit with. Developers report being slapped with the same label for innocent code review requests, then hitting the "ineligible" wall on identity verification. So this isn't a random bug. It's an automated scan that OpenAI knows about internally, with no documentation. No criteria, no functioning appeal mechanism. Another detail. ChatGPT and the API behave as if they're a single account. Get flagged on one and the other goes down at the same time. It takes out anyone who built production on top of OpenAI from a single point of failure. Tomorrow it could be anyone's turn.

@theUBIguy I expect plumbing, welding, and carpentry will remain human jobs even after it’s obvious ASIs do all intellectual work and generally run the world. You may need to wear AR glasses and become a meat puppet but I don’t believe a robot will be able to install a toilet any time soon.

@goodside So what 2 year reskilling program should I take?

AI will take some jobs, but it will create countless new jobs too—exciting jobs we can’t even imagine yet. A year later those will also be done by AI, but there will be new jobs—exciting jobs we can’t even imagine yet. Six months later those too will be done by AI, but

@viian_iv It’s unlikely anyone will heed your call, so you may need to murder me yourself. Let me know if you need any help facilitating—my DMs are open.

@goodside and my original statement remains!

@viian_iv This conversation started with you joking I should be shot. You can misconstrue the above, but I think you know what I meant—that I’ve had an unforeseeable negative effect on your mood. If you understood my remark as sexual, that was emphatically not my intent. I wish you well.

@goodside rape-y too, y'all are fucking weirdos man

@viian_iv Undoubtedly, then, I have touched you.

@goodside doing the thing. twitter is an example of how parasites like you ruin everything you touch

This is a joke for like 10 people but I stand by it. Shout out to everyone who read The Mismeasure of Man and hated it—you’re the real OGs.

It’d be funnier if Dawkins hated LLMs because then his nemesis, depending on the year, would be Gould, God, or Claude.

@scottsantens I usually limit this account to a mix of entertainment and opinion as I consider advocacy somewhat over-crowded but I’ll look into it more.

@goodside Fair enough. You could share it with others though who may feel differently.

@Orwelian84 Oh my science.

@goodside i suspect that when the first proper AI churches get going the next "new wave" athiest movement will have that exact rotating fight. wait till AI-Jesus has debates with Ai-Hitchens.

@giffmana FWIW I think that was the right call. I often delete tweets with no justification beyond them attracting too little engagement. I think people should be less afraid of deleting in general.

@goodside Yeah realized shortly after. But... if i edit my post, the comments pointing it out disappear. I guess I'll just delete instead.

@pbwinston I agree the cognitive top percentiles may be fine for a decade or more, but what new jobs do you imagine exist outside this sphere for a person currently employed at McDonald’s, with, say, 10 years prior experience doing data entry work?

@goodside My mental model: imagine a sphere, these are jobs that are automated. Humans take jobs outside that sphere. The sphere is always growing, but there always space outside the sphere, because the sphere is in empty space, not inside a room it can fill.

Touché. For context, I tend to delete tweets that attract very little engagement so my timeline is more interesting to passersby. My deleted tweet (which I otherwise stand by) read: > Ricardo’s theory of comparative advantage implies ASI gods on a Dyson sphere slowly engulfing Sol would still value us as trading partners as much as it implies we would buy fruit from chimpanzees instead of foraging for it ourselves—i.e., not at all.

Response to @goodside’s deleted tweet: honeybees.

@scottsantens I do think something like UBI or dividends may be needed but I don’t have a considered enough opinion on the issue to be signing pledges right now.

@goodside Will you sign the AI Pledge for Humanity in support of everyone benefiting from AI via a dividend? actionnetwork.org/petitions/the-…

@ciphergoth I think this is an XKCD 2501 for us. My joke is mocking an archetype of an optimistic AI hype poster that isn’t familiar to people outside our corner of X. A typical SNL audience would not laugh at my joke.

@goodside I'm sorry, it's extremely obvious and anyone who failed to see it only has themselves to blame.

@AndreTI Yeah. The specific problem is someone famous quote-tweeted it to an audience far enough removed from AI twitter to be confused; my reply was mostly for that audience.

@goodside I don't feel like this was your fault. The joke was extremely straightforward. There are just angry people on here with very poor reading comprehension.