Michael 🔸

Executive Director, @aisafetyanz | Let's make sure this goes well, alright?

[Michael 🔸 @mjkerrison]

Let's tackle poverty, animal suffering, and the disenfranchisement of future generations!

[Dr. Peter Slattery @PeterSlattery1]

Who is most vulnerable to risk from AI? And who is most responsible for addressing them? In our three-round Delphi study with 272 AI experts, we found a clear AI responsibility gap.

Experts judged that AI users and affected stakeholders are often the most vulnerable to AI risks. But they assigned primary responsibility for reducing those risks to general-purpose AI developers and governance actors, including governments, regulators, and standards bodies.

That matters because the people most exposed to AI harms often have the least power to prevent them.

💡 A few other key findings:

1️⃣ Under business as usual, experts assigned a ≥10% probability of catastrophic outcomes to 18 of 24 AI risk domains over the next 5 years.

2️⃣ Even assuming pragmatic mitigations, 5 risks remained above the 10% catastrophic threshold: dangerous AI capabilities, cyberattacks and weapons, environmental harm, inequality, and power centralization.

3️⃣ Information, finance, and national security were rated the sectors most vulnerable to AI risks.

🔗How can you engage? See our (fancy) new webpage for our interactive summaries of the findings and preprint, and please share with anyone working on AI risk, governance, or policy (links in comments).

This research is part of the MIT AI Risk Initiative (@MITAIRisk), which aims to help society understand, prioritize, and manage risks from AI. The initiative includes the MIT AI Risk Repository, a living database of more than 1,700 AI risks, the AI Incident Tracker, a collaboration with the Responsible AI Collaborative, which connects risks to over 1,400 incidents, and the MIT AI Governance Map, which analyzes risk coverage across more than 1,000 laws, standards, policies, and other governance documents curated by the Center for Security and Emerging Technology (CSET).

This work was led by Alexander Saeri, Jess Graham, and Michael Noetel (@mnoetel), with a lot of feedback and support from Neil Thompson (@ProfNeilT) at MIT FutureTech (@MITFutureTech) and MIT Sloan @MITSloan. Thanks to the 272 participants, who very generously contributed their expertise to make the findings possible.

Webpage: https://t.co/qPsLHIuJGh
Paper: https://t.co/Vcmx7N4B8b

[Michael 🔸 @mjkerrison]

@quetzal_rainbow I think this is right and I think the cyborgists know it and I think they're doing it deliberately

[Michael 🔸 @mjkerrison]

Accidentally spent $50 researching options to selectively rebind keys on Fedora...

[Michael 🔸 @mjkerrison]

@appelbolt @PlastiqSoldier @ben_golub Which is basically what you're saying but one more specific kind

[Michael 🔸 @mjkerrison]

@appelbolt @PlastiqSoldier @ben_golub or treacherous turns could be mistimed before full independence

[Michael 🔸 @mjkerrison]

Iain M Banks writing drone dialogue

[QC @QiaochuYuan]

the models write more interesting stuff when they're not pretending to be some guy. they are not some guy. this is how fable sounds when it's writing tweets while pretending to be itself

[Michael 🔸 @mjkerrison]

@MatriceJacobine It's treatlerism all over again

[Michael 🔸 @mjkerrison]

@sightspinner @codytfenwick Incorrect

[Michael 🔸 @mjkerrison]

@CharlieBull0ck I think a companion to this is an observation about much of the anti-safety discourse feeling rooted in contingent facts and culture around software and the internet specifically. Wonder if it would play differently if LLMs happened to be built out of Soviet water computers.

[port @smkmnstr]

@AndyMasley Someone fixed it, can’t find where I saw it originally (on here)

[Michael 🔸 @mjkerrison]

@smkmnstr @AndyMasley I laughed out loud at this one

[Michael 🔸 @mjkerrison]

@LeRoyDesCimes In my head I'm pronouncing it as a glottal stop

[Michael 🔸 @mjkerrison]

@danielmurfet Wow! Are you still enthusiastic about Timaeus and its agenda as well?

[Michael 🔸 @mjkerrison]

@robertwiblin I literally never see anyone ask "and what if that's right", but am not really surprised

[Michael 🔸 @mjkerrison]

@zetalyrae ACX's, both Scott's and the contest finalists'. Still think about Albion's Seed and Cities and the Wealth of Nations.

[Michael 🔸 @mjkerrison]

@sethlazar Who ought to be making those two decisions instead, why do you think they aren't (descriptively), and how much does that hinge on whether you buy arguments about particular risks?

[Michael 🔸 @mjkerrison]

@TheZvi Wait, I haven't seen this - what are people talking about?

[Apollo Research @apolloaievals]

We're hiring. If you want to work on this and even more interesting work on scheming (not yet public), please apply!

[Michael 🔸 @mjkerrison]

@ariel_zj Exciting - currently just working through them in release order (so on Use of Weapons now)

[Michael 🔸 @mjkerrison]

Wow Player of Games is excellent