Search results

They just don't get it, do they? Computers are supposed to be deterministic. It's in the very name: they compute, that's a deterministic process. I know benchmarks are used routinely in machine learning, although academics getting published because they bumped the accuracy a few percentage...

Is that like one of those pairings where they charge someone for both trying/plotting to do something and then also actually successfully acting out the plan? They sound very similar to me, but obviously IANAL.

How many times do we need to yell it from the rooftops, those so-called AI products are still little more than glorified auto-complete. The recent article about RL(HF) doesn't exactly help clear things up either because it buys into the misleading analogy proposed: that even humans use imitation...

I thought Twitter was supposed to perish within a year after Musk purchased it? Whatever happened to the predictions of its demise? If he couldn't bring it down fast enough, perhaps it's high time someone helped move it along.

What always puzzles me is, if you lose trust that vaccines work (which is a possibility given how hard it is to truly understand how most modern technologies work, medical and otherwise), why would you trust what an anti-vaxxer, or anyone really, has to say about it? Why is the default reaction...

That and in the case of CSAM in particular, it also seems to be that they tend to go after the big fish first/mainly, the guys churning tens of thousands of files through a wider network. I don't know if that's because those are the most news-worthy, the easiest or most valuable targets for law...

For me, apart from the obvious and persistent reliability issue, one other major issue with the proposition of outsourcing generation of problem solutions to AI is that it implicitly seeks to relegate humans to the position of verifier. We see this with use cases such as generating various kinds...

What, you mean this? I was honestly expecting a bit more in-depth analysis or some kind of description of what is now involved in debugging the output of these models.

Do we have a video (schematic or real microscopy) anywhere of bacteria battling cells using syringes? I found this basic animation about how P. syringae infect plants but no clear syringe action... I'm just fascinated by the weird and wonderful world of molecular technology that single-celled...

So we've all heard about multiple energy levels of electrons, which is simple enough to teach it to high schoolers (at least when I was there). Transmon doesn't offer the same intuitiveness for me, so I wonder what other phenomena could be used to create general quantum digits? Maybe something...

The tech bros push out more vaguely named models that are all slightly different so the researchers are always a step or two behind the latest and greatest and they can just say, but the research that shows minor improvements was only for the previous model, this one is better.

I'd bet my bottom dollar the amount of testing and legislation in place to make cars safe vastly exceeds the amount used to ensure LLM safety and efficacy. Once the two approach each other, then maybe you can use that analogy.

Should have just used /s I guess.

It's just as well, dollars won't be worth much soon anyway if things keep going as they are?

Where's Anonymous when you need them?

Since the image has a high enough resolution, zooming the page works too.

If they just benchmark it on math questions, the answers to those are verifiably correct or incorrect, and if they have a magic knob that you can use to reduce errors by a vague amount, how does that help guarantee the answer to questions like that is usable? Of course, trying to come up with a...

So I've heard, but would you like to cite some hard data to back up the claim? I bought one of the newest generation ones recently as a gift, it wasn't for the intelligence bit but because the person's old iPhone was like 10 years old... for whatever that's worth.

Yea that's all well and good, but can I ask one question before you go on? Whatever happened to all those allegations that most or all of the training data for your LLMs was not ethically obtained or outright stolen? Shouldn't there be some restraint before warping everything around a thing with...

Who knew telescreens would turn out to be so expensive?