mlyle

20 ms·

391.
by mlyle 1y ago
I don't really like your snarky initial take, even when I'm the guy calling bullshit in the first place in the thread. People propagate falsehoods for numerous reasons. The first is, they don't know it's false. They he
392.
by mlyle 1y ago
Bullshit. There is no way that living things are releasing enough ionizing radiation to interfere with a computer, especially an older one-- attenuated both by the rest of their flesh, the building's walls, the computer's chassis
393.
by mlyle 1y ago
How much of it doesn't? We're deterministic? (aren't we less deterministic than LLMs?) All of our training is auditable? (there's a wealth of unknown experiences in each person writing code, to say nothing of the unknow
394.
by mlyle 1y ago
It's usually not the same pile of meat defining the problem and solving the problem. Yes, humans exceed the capability of machines, until they don't. Machines exceed humans in more and more domains. The style of argument you made
395.
by mlyle 1y ago
Nah-- I feel like I have my eyes pretty wide open about the shortcomings of LLMs (but still find them useful often). But any argument seeking to dunk on LLMs needs to not also apply equally to the alternative (humans).
396.
by mlyle 1y ago
Nearly as bad: trying to use systems made out of meat, evolved from a unrelated background and trained on an undocumented and chaotic corpus of data, to try and produce content satisfying a strict mathematical model.
397.
by mlyle 1y ago
I've not read the details, but given that things aren't being matched against themselves-- Seems like you need to send ~ 2^16 requests and then ~ 2^16 spoofed replies.
398.
by mlyle 1y ago
This is really great. Bummed I didn't see it before. I made a really precise timebase and a fairly precise interval counter a decade ago using pretty different techniques (ECL and staggered delay lines and a lot of calibration). It w
399.
by mlyle 1y ago
Usually we need evidence to make a decent decision. It's not debate club.
400.
by mlyle 1y ago
The deal is supposed to be that there's a trade: * You are given an exclusive right to exploit a work, for enough time to make it worth your while. * Everyone gets the work in the end. We're not succeeding at this. The terms are
401.
by mlyle 1y ago
Facebook gets a fiscal benefit when the counterparty to the contract breaks the rule, and so has no incentive to enforce it (rather, the opposite). Unless, of course, Facebook is held accountable for not enforcing it.
402.
by mlyle 1y ago
I love the old HP app note linked: https://ilrs.gsfc.nasa.gov/docs/time_interval_measurements.p... The fact that we have this discussed in the open, in such a transparent way, is great. HP was confident they could mai
403.
by mlyle 1y ago
The problem is, the opposite approach is... "We're scot free, because we told *wink* people to not sell us sensitive data. We get the benefit from it, and we make it really easy for people to sign up and get paid to give us this
404.
by mlyle 1y ago
> that learning is a linear process where you master one topic and move on to the next. It's closer to true in mathematics than most other places, but not very close to true . It's amazing the "layer cake model" of m
405.
by mlyle 1y ago
> You can do that manually. Absolutely. You can spend time on figuring out what to do next, and how, and how to do spaced repetition for the material and test yourself effectively. There are aspects you'll do better than a set cur
406.
by mlyle 1y ago
IMO it's scaffolded and explained a bit more than an average mathematics lesson, though teachers vary a lot. There's a whole lot of "here's the formula" and not so much "here's the derivation" in most
407.
by mlyle 1y ago
> but it beats paying $50 per month for the service OP's post links to. I used this with my kids. The $50 a month to make it easy is worth it for many people, and is downright cheap compared to many approaches. Now, there's oth
408.
by mlyle 1y ago
I think it's interesting to think about, and still somewhat uncertain: * How much a large language model is effectively a world model (indeed, language tries to model the world...)? * How much do humans use language in their modeling a
409.
by mlyle 1y ago
So, I guess it was just a low effort comment, and I shouldn't take you seriously? I thought you were making an actual effort to converse. Generating tokens to represent steps isn't just "cosplaying" --- it's repres
410.
by mlyle 1y ago
So, I'm not sure 180 IQ intelligence can. But something that has equivalence to high human intelligence (say 140IQ, generalized to a wide variety of domains) and can engage in long-term directed, purposeful, and autonomous behavior w
411.
by mlyle 1y ago
I would argue that AI is able to do a lot of what a 110 IQ human does today already-- at least the 30% of the population that is of average intelligence and has a below average appetite for growth and challenge. And sure, it's not perf
412.
by mlyle 1y ago
> I'm eagerly waiting for the next big breakthrough when someone trains a model that switches between "thinking" and "outputting" modes mid-answer :) ??? That was last year's breakthrough.
413.
by mlyle 1y ago
To me, GPT-5 feels slightly better. And rumor is the cost for OpenAI to provide it is much less.
414.
by mlyle 1y ago
> The reason why true singularity-level ASI will never happen isn't because the technology cannot support it (though, to be clear: It can't). Its because no one actually wants it. Plenty of people want something capable of doin
415.
by mlyle 1y ago
Here there's weirdness, still, though because there's such a frequency difference you'll get between "low priority" and "high priority" cores.
416.
by mlyle 1y ago
> There is a wider point that ChatGPT is less autonomous than an assistant, as no matter the tenure with it, you'll not give it the level of autonomy that a human assistant would have as it would self correct to a level where you&#x
417.
by mlyle 1y ago
> It is absolutely impossible that human assistants being given those tasks would use even remotely within the same order of magnitude the power that LLM’s use. A human eats 2000 kilocalories of food per day. Thus, sitting around for an
418.
by mlyle 1y ago
> When doing automations that are perfectly handled by deterministic systems why would I put the outcomes of those in the hands of a non-deterministic one? The stuff I'm punting isn't stuff I can automate. It's stuff like
419.
by mlyle 1y ago
I think this is pretty silly either way. - There's an upstream loss on electricity directly in proportion to how much you use; ignoring this tilts the analysis in favor of electricity. - You pay more for heat from electricity than gas,
420.
by mlyle 1y ago
Sure, it's "equally efficient" if you ignore the inefficient thing that is done outside where you draw the system box, directly in proportion to how much you do it. Heating my house with a giant diesel-powered radiant heater

More ›