lamename

You _can_ but you should not.

Yes you put in a "probe" (fiber optic), just like when you stimulate with an electrode. The difference is electricity stimulates everything, while the light can be targeted, not by what it illuminates but what responds* because...

Ahead of time, you either use viral tools or other genetic modifications to make specific cell types express the photorecrptor.

*there is an effect of light/heat that all cells are subject to, and good experiments try to control for this as best they can

I mostly agree with everything you said. Do you feel the same way about code written by an LLM?

Exactly. HN darling Paul Graham writes this way.

I find the constant critique of punchy style a bit tiring. It would be more productive for the grandparent to think about the content and state an opinion.

The link in the article that is right near the words you're talking about links to a wikipedia page that says the book is from 2005. So I conclude it was 2005 or soon after

Do you disagree with the point made?

Not quite, but these help

https://poloclub.github.io/transformer-explainer/

https://youtu.be/wjZofJX0v4M?si=gT8Zlz1IY14KV_ju

So far i really like what it does for the example articles shown. I want to test it on 1 or 2 articles I know well, and if it passes that test it's a product I'd totally pay for.

I tried to upload a 239 KB pdf and it said "Daily processing limit reached".

I generally agree with the broader point you're making, but I also think there's nothing wrong with pointing out how messed up it is that that's the reality of the choice. The whole point of improving society is to eliminate this kind of dilemma

Maybe. Those could be the case. But ignoring all confounding factors, this phenomenon is possible with numerical experiments alone. One of the meanings of "the Law of Small Numbers".

Basically, the possibility that the small study was underpowered, and just lucky...then the large studies with more power are closer to the truth. https://en.wikipedia.org/wiki/Faulty_generalization

I agree with most everything you said. The problem has always been the short-term job loss, particularly today where society as a whole has resources for safety nets, but hasnt implemented them.

Anger at companies who hold power in multiple places to prevent and worsen this situation for people is valid anger.

As much as I like the article, I begrudgingly agree with you, which is why I think the author mentions the physical constraints of energy as the future wall that companies will have to deal with.

The question is do we think that will actually happen?

Personally I would love if it did, then this post would have the last laugh (as would I), but I think companies realize this energy problem already. Just search for the headlines of big tech funding or otherwise supporting nuclear reactors, power grid upgrades, etc.

In my experience in neuroscience it even differs widely across programs/universities. Some good professors care about giving good talks, and if you're lucky it becomes contagious in the program. Others think less of you if it's clear, some are too naive to realize obscurity is not a virtue.

Yeah, but still "scary" because you have to be really careful to not fool yourself and pay attention even with those algorithms. For example, a good demonstration with tsne https://distill.pub/2016/misread-tsne/?hl=cs

Being there 24/7? Yes. Better job? I'll believe it when I see it. You're arguing 2 different things at once

Wow a sane person among all the hype. Great to see you!

I really do agree with your point overall, but in a technical paper I do think even word choice can be implicitly a claim. Scientists present what they know or are claiming and thus word it carefully.

My background is neuroscience, where anthropomorphising is particularly discouraged, because it assumes knowledge or certainty of an unknowable internal state, so the language is carefully constructed e.g. when explaining animal behavior, and it's for good reason.

I think the same is true here for a model "knowing" somethig, both in isolation within this paper, and come on, consider the broader context of AI and AGI as a whole. Thus it's the responsibility of the authors to write accordingly. If it were a blog I wouldn't care, but it's not. I hold technical papers to a higher standard.

If we simply disagree that's fine, but we do disagree.

I agree with your point except for scientific papers. Let's push ourselves to use precise, non-shorthand or hand waving in technical papers and publications, yes? If not there, of all places, then where?

I agree. Isn't this just utilizing the representation learning that's happened under the hood of the LLM?

Have you seen the statistics about high impact journals having higher retraction/unverified rates on papers?

The root causes can be argued...but keep that in mind.

No single paper is proof. Bodies of work across many labs, independent verification, etc is the actual gold standard.

The Bullshit asymmetry principle comes to mind https://en.wikipedia.org/wiki/Brandolini%27s_law

Adjusted for inflation? Without (crippling) debt accrual and adequate emergency fund, retirement, etc? Did you have children or childcare expenses? These all knock on that total compensation quickly these days, which is the main argument in this particular thread of replies.

We agree, minimum wage doesnt mean that. And in a large metro area, that's why $120k is closer to min wage than a good standard of lliving and building retirement.

I disagree. Your data doesnt make the grandparent's assertion false. Cost of living != per capita or median income. Factoring in sensible retirement, expensive housing, inflation, etc, I think the $120k figure may not be perfect, but is close enough to reality.

Have you ever noticed that stocks can go up as enshittification also goes up?

It's simpler than that. "Prestigious" universities emphasize research prestige over all else on faculty. Faculty optimize for it and some even delight in being "hard" (bad) teachers because they see it as beneath them.

Less "prestigious" universities apply less of that pressure.

Yes I agree. They will and should blame the human. That's a problem when the human isnt given enough time to complete projects because "AI is SO productive"

> AI is always being touted as the tool to replace the other guy's job. But in reality it only appears to do a good job because you don't understand the other guy's job.

This is a well considered point that not enough of us admit. Yes many jobs are rote or repetitive, but many more jobs, of all flavors, done well have subtleties that will be lost when things are automated. And no I do not think that some "80% done by AI is good enough" because errors propagate through a system (even if that system is a company or society), AND the people evaluating that "good enough" are not necessarily going to be those experienced in that same domain.

They will demand use of AI tools for "productivity" and then complain when there are bugs in prod without realizing the root cause.