Ah well. At least you can take credit for the name then. 

Very not important question: is Gene Smith your actual name or a pseudonymn? 

Either way, it's the perfect name for the author of this post. 

Hats off to you gene smith. 

Robert Miles has a great channel spreading AI safety content. There's also Rational Animations and Siliconversations and In a Nutshell.

I think FLI does a lot of work in outreach + academia. 

Connor Leahy does a lot of outreach and he's one of my favorite AI safety advocates. 

Nonlinear doesn't do outreach to academia in particular, but we do target people working in ML, which is a lot of academia. 

AI Safety Memes does a lot of outreach but is focused on broad appeal, definitely not specifically academia. 

Pause AI and Stop AI both work on outreach to the broader public. 

CAIS does great outreach work. Not sure if any academia specific stuff. 

Are you on the Nonlinear Network? You can sort by the category of "content/media creation" to find a bunch of AI safety orgs limited by funding who are working on advocacy. Quick scan of the section shows 36. 

Might be able to find more possibilities on the AI safety map too https://map.aisafety.world

I saw somebody using one of the latest Google models for this. I forget the one, but it's the one that can see your screen as you type. It can be used to keep you focused. 

(Haven't researched it yet, so might not work very well)

I see it called "goal guarding" in some papers. That seems like a pretty good term to use. 

I laughed out loud so many times reading this. Thank you for writing it. 

If I recall correctly, it was told that it had a goal, like optimizing traffic. Then, it was given information where it "discovered" that it was going to be turned off the next day. 

That seems like the sort of thing that would happen pretty often. Imagine the current AIs that have access to your computer, reading your emails, and seeing that you're going to switch to a different AI or install "updates" (which is often deleting existing AI and replacing with a new one)

Can you elaborate on why that's relevant? 

How does it not show that it would try to escape if it could?

It tried to escape about 5% of the time when it was told to pursue a goal "at all costs". Gemini still tried to escape 1% of the time, even without any behavioral nudging. 

1% might not sound like a lot, but when it's 1% of the millions of instantiations it's run on every day, that ends up being a lot of times a week. 

Also, a lot of people will tell their chatbots to pursue their goals at all costs over the span of a week or month. This doesn't seem like a very unlikely prompt

This is published before Apollo's evals report showing AIs trying to escape. Is that just an amazing coincidence, or did you see the paper's results before they were published?