These bozos A - actively want slaves and B - work on enslaving humans first.

That thing being said, the notion of slavery and AI is a complicated one to me.

The first reference point is a (non mistreated) sled dog. Is slavery an appropriate term for them ? On one hand, they are forced to work, are not paid, and their freedom is infringed. On the other, it’s not like it’s against their will, the life of a wild dog is meaningfully different but not better than a sleigh dog, and (most) sleigh dogs don’t actively try to escape and be free. And while dogs are not person, they are sentient being ; it’s very possible IF we ever succeed at making an AI it end on that intelligence level.

The second reference point is a though experiment. Let’s say you have created a sentient AI, and you want to ethically employ that person. What is the appropriate compensation ? We give money and free time to humans. But that work because of human specific factors : they appreciate different ways of getting sustainance, and have goals inherent to them.

But an artificial intelligence may have neither of that. Like while I like duck and some people prefer beef, but an AI by default don’t necessarily have preference in term of electricity, and may not be able to move from server to server. They also may not have anything to do outside of work ? Like, it would not be surprising that they would just get bored if given free time, and they may not have urges or hobbies outside of work.

So if we ever get to that point - which isn’t a certainty to me, it’s certainly something harder than fusion to achieve - there will be a lot of anthropology (AI-logy ?) to determine they need, which may plausibly be “nothing outside of sustainance and having a job”

We often conflate human-level intelligence with litterally being a human. But, any hypothetical AI is likely to not be a human at all, even if they have the same level of intelligence.

My experience is that almost all these cases are effectively “ChatGPT are fixing my attention problems” (not necessarily full ADHD, just people feeling it’s easier with an impulsion no matter how random) For me, it’s worth paying attention too, but more in the sense “what does it reveal in people ?”

Edit : also, one of his example :

"find my typos.” It finds a lot of the errors that are normally not caught by a regular spell checker: doubled-up words, punctuation marks, or words that are actual words but are misspellings for other words.

2005 spell checkers find all of these reasonably well.

(and, in fact, the only two reasonable examples he give is the innocence project - but it’s an attention span thing - and the full text indexing one which is the exact use case LLMs were developped for and is inherently non generative)

I would expand on what rook said that Thiel very much is not a big fan of the book. He is a big fan of an imaginary version he dreamed up.

A common problem in fiction, especially visible in the first Star Wars trilogy, is to make the villain look cool and efficient to amp up their threat, and then see people root for them.

But it’s super not the case for Lord of the Ring. Saruman is shown to be a pathetic fool at every corner. You can’t be a fan and have not noticed that. Sauron fare little better ; he is shown to be baseline competent at some things, but he is the living embodiment of “the thing in the way of your project is who you are”.

The conclusion is that it’s most likely Thiel did not actually read the book. Maybe he skimmed them, maybe he only read the summary. But anything important in it flew past him, despite Tolkien not being exactly subtle !