Conversation

Excellent essay by my dear NYU colleague Julian about the existential threats from rogue AI, or rather, about the insanely-overstated likelihood of such threats. 1/
Quote Tweet
Is Elden Ring an existential risk to humanity? That is the question I consider in my new blog post. Yes, it is a comment on the superintelligence/x-risk debate. Read the full thing here: togelius.blogspot.com/2023/04/is-eld Or follow along for a few tweets first, if you prefer.
Show this thread
Image
45
590
Julian reminds us: 1. all intelligence is specialized, including human intelligence. 2. being smart in some domains makes you strong in some environments but weak in others. 3. Intelligence does not immediately cause a thing to be able to "take over" 2/
4. Intelligence does not immediately cause an entity yo want to "take over" 5. A very dumb but specialized entity can kill a smarter one, e.g. virus vs human. Julian argues that MS Excel can do intelligent tasks and, in some ways, has already "taken over" our lives 🤣🔥
25
269
Inappropriate application of AI could very likely cause unwanted disruptions in the economy or necessary government oversight mechanisms. A commission should be formed to address this proactively. ✌️
1
"All intelligence is specialised". Someone finally did it. They did the math. (AGI as a statement is fine, in practice it is a specialist in action, attempting to generalise)
2
I don't know what #1/2 is supposed to mean, seems clearly wrong to me. Human can build machines/tools that outperform any animal in any domain. Crows/chimps may have better memory, but they need to be protected to avoid being killed by us, there's a fundamental difference here.
1
There are a lot of nice allegories in this article . Sadly the argument of self improving AI is mentioned , but not addressed. An AI, which optimizes infinitely with for a certain goal, finds interesting shortcuts, deconstructs all atoms for it.
Yeah, but being specialized in one domain (survival in the African savanna) got us to the Moon and beyond. Even if AGI can’t do “everything” it still has the potential to … do a lot
AGI needs to become really good at making machines that make machines that can kill us. I’m happy to take a baseball bat to Alpha go and see who wins that “chess match”.
Show more replies