MarkusQ

12 ms·

271.
by MarkusQ 5mo ago
That's actually a fantastic idea! Oh wait, no, I'm an e-addict. Drat! Curse this monkey!
272.
by MarkusQ 5mo ago
Nuts. You're saying the discovery that humans can process language without being conscious "couldn't possibly" inform the debate about LLMs? When that debate is literal predicated on the assumption that the ability to p
273.
by MarkusQ 5mo ago
You are dangerously close to initiating a new sport. Someday, commentators filling air time on ESPN's "Need to Yeet" may bring up this comment as "the Casual Remark That Started it All".
274.
by MarkusQ 5mo ago
If by "attacked" you mean "repelled attacks by," then sure.
275.
by MarkusQ 5mo ago
Right. Iran attacked US ships in international waters. I guess you could say the "first step" was the the ships being there to be shot at, if you were trying really, really hard to spin it as the united states being the aggressor
276.
by MarkusQ 5mo ago
"Iran launched missiles, drones and small-boat attacks at U.S. warships near the Strait of Hormuz, and that the U.S. responded by intercepting the threats and striking Iranian military sites responsible for the attacks." Iran atta
277.
by MarkusQ 5mo ago
Tesla Model3 dual is 4.03 seconds, not 4.3 while the F-150 lightning standard range is 4.2. The F-150 extended range is 3.8 as you state, but then the Tesla Model 3 performance comes in at 2.8. https://www.0-60specs.com/tesl
278.
by MarkusQ 5mo ago
A quick back of the envelope calculation shows that airborne microplastics can't possibly be significantly contributing to global warming. That's not surprising; there are millions of other things that aren't contributing to
279.
by MarkusQ 5mo ago
> The growth of AI feels a little like losing a limb Or gaining a new, oddly misshapen and inexplicably placed limb of no apparent purpose or utility.
280.
by MarkusQ 5mo ago
See my response there as well.
281.
by MarkusQ 5mo ago
It is as (or more) common for that type of construct to be used to set up tension for subsequent exploration. "Can light really be both particles and waves?" I also find it interesting that the "Dawkins is clueless" arg
282.
by MarkusQ 5mo ago
What? How does their saying that the study is is far from conclusive "badly undermine" my point?
283.
by MarkusQ 5mo ago
I am 100% in favor of real research, basic or otherwise. That's _why_ I'm so opposed to hype, grift, misrepresentation of results, p-hacking, and all the other nonsense. Science is measured in explanatory progress and facts disco
284.
by MarkusQ 5mo ago
> The warming impact is tiny, far less than the impact of carbon dioxide emissions, and only a fraction of the impact of soot. So...they may be doing something. Or not. But if they are, it's not much. More study is needed to keep
285.
by MarkusQ 5mo ago
Obviously, there is none. The point is, assuming the answer to be "yes" isn't a slam-dunk, and (in general) may not even be a good bet. (The mechanical reason for Betteridge being true more often than not is that if a journa
286.
by MarkusQ 5mo ago
Citation needed. All of the direct quotes I've seen have clearly stated that he can not _disprove_ the claim of consciousness, and finds this fact interesting.
287.
by MarkusQ 5mo ago
TIL that my mind has dynamic memory and enjoys an occasional refresh cycle. Thank you!
288.
by MarkusQ 5mo ago
>> "Dawkins did not proclaim Claude conscious" > This is true in the literal sense that Dawkins didn't explicitly say "Claude is conscious" It is true not only in the literal sense, but in the rhetorical se
289.
by MarkusQ 5mo ago
That's a good example of my point about reading comprehension. The headline is "When Dawkins met Claude Could this AI be conscious?". That's a question, not a statement. By Betteridge's Law of Headlines, which sta
290.
by MarkusQ 5mo ago
The lack of reading comprehension (or perhaps just lack of reading) behind this brouhaha is amazing. Dawkins did not proclaim Claude conscious. He argued that Claude passes the Turing test, and then asks a question: if something can pass t
291.
by MarkusQ 6mo ago
"the model can alter its priors" The model doesn't have high-level priors in the Bayesian sense (though you could have priors about it). The low-level priors it does have (the weights) are not modified by the context.
292.
by MarkusQ 6mo ago
Yeah. IMHO this is the more serious objection.
293.
by MarkusQ 6mo ago
Close famously counts in horseshoes and hand grenades. Algorithms, just as famously, are a domain where off-by-one is still wrong.
294.
by MarkusQ 6mo ago
So substitute another phrase, if you prefer. It doesn't change the logic. "Specifically, we define a formal world where bungling is defined as inconsistencies between a computable LLM and a computable ground truth function. By em
295.
by MarkusQ 6mo ago
Except that latent space does not change in response to new information, something that thoughts famously do. If you read a book that captures the author's thoughts, disagree, and write an eloquent arguments to the author, you might c
296.
by MarkusQ 6mo ago
That would be "a profitable AI company that isn't just fudging their numbers." I do not have an example to offer.
297.
by MarkusQ 6mo ago
Being untrue is what makes these things lies[1][2]. Arguing that they are emotionally compelling is just confirming that they're manipulative lies. [1] https://www.in2013dollars.com/New-cars/price-inflation [2] h
298.
by MarkusQ 6mo ago
It _increases_ the signal to noise ratio. It's a denoising technique, and that's what they do. Compared to the processing already done to get data from astronomical data, yeah, it's essentially free.
299.
by MarkusQ 6mo ago
Sure. Lying to people and telling them they're somehow oppressed just because other people are more successful than they are can be used to stir up a revolution and get a lot of the useful idiots killed. That's why it's impo
300.
by MarkusQ 6mo ago
The main claim that I was claiming that TFA the claims to be incorrectly claimed to be impossible by the LLM-hype critic is roughly "current generative AI business models can be successful". The critic initially argued that there&

More ›