HN user

jhrmnn

927 karma

Affiliated to Microsoft Research

https://hrmnn.net

[ my public key: https://keybase.io/jhrmnn; my proof: https://keybase.io/jhrmnn/sigs/DZ6d_BcOm6y49gxqGhsmSZ8w_BonUt3q5JXjpd8GmRU ]

Posts3
Comments296
View on HN
Claude Tag 29 days ago

Reliability is one property of a product. And it’s quantitative, not qualitative, so it all comes down to trade-offs

When no work is safe from mechanization, surely the value of labor wrt capital must fall, and the societal pressure on redistribution will rise. The ultimate outcome of technological progress is either extreme inequality or massive redistribution

Nothing, it’s that same story again. Industrialization turned peasants to blue collar workers by mechanizing agriculture. Then blue collar workers were turned to white collar workers by mechanizing all manual labor. Now AI is coming for white collar workers by mechanizing intellectual labor. The big question is what will white collar workers turn into.

So most training data would be grey and a little bit coloured? Ok, that sounds plausible. But then maybe they tried and the current models get it already right 99.99% of the time, so observing any improvement is very hard.

Interesting. The debate about whether the artist matters in perceiving a piece of art is very old. You don’t seem to consider the possibility that the artist’s intent matters when listening to music. For me it absolutely does. As the AI has no intent (agency), the AI music is void of any value to me.

Think of a woodworking project. Compare doing everything old-school by hand vs using modern tools to go faster. Think about the end product being just an item with a function vs it having some design value or even craftsmanship value. Does the parallel work?

But individual behavior is not about preventing climate change, it’s about doing what’s right. It’s wrong to pollute the environment, one way or another. A single person not stealing won’t reduce the crime rate in a country yet it is the right thing to do.

This depends very much on what "practical purposes" are. For almost all conceivable technology, relativistic quantum mechanics for electrons and light, ie QED, is sufficient fundamental theory. This is unlike before quantum mechanics, when we basically didn't have fundamental laws for chemistry and solid-state physics.

I’m guessing what was meant is that the price of things that are to be invested in is growing wrt the price of things that are to be consumed. Which naively makes sense to me in an economy based on growth where the total consumption starts to stagnate—the surplus still has to go somewhere. Is it so or is reality more complicated than that?

I think it should have been “If you need oxygen and have a CNS, then you need sleep.” Other tissues can take oxidative break during wakefulness, but since CNS is _generating_ wakefulness, if it takes a break, by construction there is sleep.

useless ritual fluff

I believe that LLMs can be very useful to identify this stuff in our processes. The solution shouldn't be then to fill them with LLMs but strip them entirely away. I tend to think the same about everyone freaking out about LLM misuse in education.