[-] [email protected] 1 points 4 days ago

Gotcha. Thanks for the explanation!

[-] [email protected] 1 points 4 days ago

If he has now got supreme power couldn't he just declare that the president doesn't have supreme power, and hey presto it's gone?

[-] [email protected] 3 points 1 month ago

Maybe I misunderstood the OP? Idk

[-] [email protected] 13 points 1 month ago* (last edited 1 month ago)

People sometimes act like the models can only reproduce their training data, which is what I'm saying is wrong. They do generalise.

During training the models are trained to predict the next word, but after training the network is always effectively interpolating between the training examples it has memorised. But this interpolation doesn't happen in text space but in a very high dimensional abstract semantic representation space, a 'concept space'.

Now imagine that you have memorised two paragraphs that occupy two points in concept space. And then you interpolate between them. This gives you a new point, potentially unseen during training, a new concept, that is in some ways analogous to the two paragraphs you memorised, but still fundamentally different, and potentially novel.

[-] [email protected] 16 points 1 month ago

Not an ELI5, sorry. I'm an AI PhD, and I want to push back against the premises a lil bit.

Why do you assume they don't know? Like what do you mean by "know"? Are you taking about conscious subjective experience? or consistency of output? or an internal world model?

There's lots of evidence to indicate they are not conscious, although they can exhibit theory of mind. Eg: https://arxiv.org/pdf/2308.08708.pdf

For consistency of output and internal world models, however, their is mounting evidence to suggest convergence on a shared representation of reality. Eg this paper published 2 days ago: https://arxiv.org/abs/2405.07987

The idea that these models are just stochastic parrots that only probabilisticly repeat their training data isn't correct, although it is often repeated online for some reason.

A little evidence that comes to my mind is this paper showing models can understand rare English grammatical structures even if those structures are deliberately withheld during training: https://arxiv.org/abs/2403.19827

[-] [email protected] 4 points 1 month ago

Words aren't real, we can spell them however we want. Don't be weird.

[-] [email protected] 27 points 1 month ago

But at what cost!?

[-] [email protected] 3 points 2 months ago

Would a Trump victory reduce the chance of WW3?

[-] [email protected] 10 points 2 months ago

The whole misinformation angle is bullshit. It's such a lib mindset to think your enemies are just misinformed.

The majority of people at Jan 6 were business owners or upper management. They came from urban communities that had seen a decline in the white population. https://d3qi0qp55mx5f5.cloudfront.net/cpost/i/docs/Pape_AmericanInsurrectionistMovement_2022-01-02.pdf

Like every fascist movement Jan 6 was the petite bourgeois attempting to shut down democracy and install a dictator that would stop demographic changes and keep the current hierarchies unchanged despite the falling rate of profit and the resulting monopolisation / need for the growth of the working class.

But if you refuse to see it as a systemic, almost inevitable process, then you're left thinking the Jan 6 people were just CraZzyYy. Maybe it's that damn internet the kids are always on!

[-] [email protected] 0 points 2 months ago

What are hoops that you chase with a stick doing to children? Demands grow to restrict kids access to hoops that you chase with a stick

view more: next ›

HorseRabbit

joined 1 year ago