articlepan

I don't think it's that complicated. Anthropic's business model from the start was to brand themselves as the safe/responsible AI company. The whole idea was that eventually society and government would see the dangers of AI and regulate it, at which point the company which had invested the most in safety would pull ahead. The sooner Anthropic can have society and government come to this realization, the sooner they can reap the benefit of all their safety/alignment investment.

Two other factors:

- The employee base and C-suite are mostly true believers in the need for AI safety. It would be organizationally very challenging to pivot, even if financial interest was pushing that way.

- Anthropic is registered as a public benefit corporation, so that even if they IPO they don't have the fiduciary duty to shareholders that most people are familiar with for public companies. They still need massive revenue to stay solvent, but they can optimize for things other than pure profit.

+1 to Player Piano, amazing book that could hardly meet the moment better.

Title is bad, it's the first line of the abstract instead of the paper title. Speculative decoding for LLM inference was published in 2022: https://arxiv.org/abs/2211.17192

This paper seems to be an improvement to speculative decoding but I haven't read it yet.

This is great! For scale degree after chord progression exercise, I'd love a variation where the chord progression is in a minor key.

I agree with your first paragraph, but not your second. Models can still hallucinate when temperature is set to zero (aka when we always choose the highest probability token from the model's output token distribution).

In my mind, hallucination is when some aspect of the model's response should be consistent with reality but is not, and the reality-inconsistent information is not directly attributable or deducible from (mis)information in the pre-training corpus.

While hallucination can be triggered by setting the temperature high, it can also be the result of many possible deficiencies in model pre- and post- training that result in the model outputting bad token probability distributions.

Money is only a bookkeeping tool for complex societies. The aim of the owner class in a worker-less world would be accumulation of important resources to improve their lives and to trade with other owners (money would likely still be used for bookkeeping here). A wealthy resource-owner might strive to maintain a large zone of land, defended by AI weaponry, that contains various industrial/agricultural facilities producing goods and services via AI.

They would use some of the goods/services produced themselves, and also trade with other owners to live happy lives with everything they need, no workers involved.

Non-owners may let the jobless working class inhabit unwanted land, until they change their minds.

Solution for 2000-2010: "the first decade of the 2000s".

I realize this is counting by decades/centuries again, but if we just do it for the first decade/century under a larger span it's easy to read.