visimo-dino

[visimo-dino @VisimoD]

@zachware @Altimor But if their goal was just to hype their products and drive up valuation, and they had no qualms about lying, why choose to lie about how dangerous their product is? Why not only lie about how amazing it’s going to be? Surely that’d be easier and have fewer negative side effects.

[visimo-dino @VisimoD]

@MindMechanical @phl43 I see your point, but would this actually deter? Say you’re invaded again a decade later. The enemy has been occupying you for 3 years, and your prospects are (1) shack up w/ a soldier of theirs or (2) starve. Would the threat of having your hair cut off keep you from option 2?

[visimo-dino @VisimoD]

@Gianpattention @hair_harbor @phl43 It’s the job of soldiers to die for their country during war (potentially). It’s the job of civilians to work hard, keep the economy running, and do other non-combat-related things to help out. Otherwise, there would be no distinction b/w the military and civilian population.

[visimo-dino @VisimoD]

@flowersslop The efficient market hypothesis states that this is nearly impossible to do in any sort of realiable way, as any/all info the AI could analyze is already priced in due to the market’s efficiency. Of course, it’s just a hypothesis—but a fairly accurate one thus far.

[visimo-dino @VisimoD]

@flowersslop @GaryMarcus Humans can’t master it though. This is basically the premise of the efficient market hypothesis—that even people who are ostensibly experts at analyzing and predicting the movements of the market seldom outperform simply buying and holding some index funds.

[visimo-dino @VisimoD]

@TheFoolishPig @deredleritt3r @KhanSaifM DeepSeek did not invent the use of MoE in LLMs. The use of MoE in LLMs occurred well before DeepSeek even existed as a company. For example, Google’s GLaM (Generalist Language Model) is a sparse MoE LLM that dates all the way back to 2021, Mixtral is an MoE LLM from 2023, etc.

[visimo-dino @VisimoD]

@BalaamAndDonkey I’m not sure what it would mean for a universe to be rational vs irrational, but to the extent I can interpret it, I’d say the universe seems pretty “rational” thus far. It follows extremely precise rules with astonishing consistency. Is that not fairly “rational”?

[visimo-dino @VisimoD]

@TheFoolishPig @deredleritt3r @KhanSaifM I disagree; it only suggests that teams of non-celebrities can *reproduce* what Ilya did, after he did it (and make some incremental improvements). I haven’t seen any evidence that DeepSeek et al can produce the level of breakthrough that Ilya-led teams have (not yet, at least).

[visimo-dino @VisimoD]

@jeanpaulblartre @Robotbeat @growing_daniel No, they’re choosing it for logistical and economic reasons that benefit them. But I still don’t understand how the datacenter being located physically close to you makes it “ideal for it to kill you,” for reasons I gave above. Can you explain why you think that?

[visimo-dino @VisimoD]

@LKHDMartian @Tyler_A_Harper This is like 99% true (the 1% is bc the CEOs and other higher-ups do *occasionally* do some PR, though it's less common). The eternal skepticism most ppl have toward their utterances is, ironically, causing them to be LESS concerned about AI-related risks than they should be.

[visimo-dino @VisimoD]

@impoliticeye @Tyler_A_Harper They actually have promised those things. Or at least, promised that they'll do everything they can to encourage government to bring them about. Of course, the issue is that most people are rightly skeptical that any of that will actually happen.

[visimo-dino @VisimoD]

@Phillius188844 @growing_daniel The vast majority of the population may not have ever heard or read anything directly from Dario himself, but they hear the echoes of what he (and Altman, Hassabis, etc.) have said, in the mainstream news, and at the water cooler, family gatherings, and so on.

[visimo-dino @VisimoD]

@jeanpaulblartre @Robotbeat @growing_daniel That's really not at all how AI risk works, though. AI is a digital technology that primarily affects the external, physical world by communicating via networks (mainly the internet). Being physically close to a datacenter running an AI doesn't really change the risk profile.

[visimo-dino @VisimoD]

@Brehmel @GaryMarcus No idea what this post means. Not sure who person A or person B is supposed to be (I'm guessing you're supposed to be person B?), nor do I know what you mean by "AGI theory," which is both extremely vague and a term that no one uses.

[visimo-dino @VisimoD]

@AIQuickFindings @rubytwos8 Not trying to be a jerk, but all of your posts (including this one) seem to be written by AI, both on an "eyeball" level and according to Pangram. So it's ironic bc your post just now mentions the AI "slop problem."

[visimo-dino @VisimoD]

@rubytwos8 There are concentric circles of why people hate AI. The circles you mentioned exist, but I don't think they're the widest ones. I think if you zoom ALL the way out, the most common reasons for ordinary ppl are job loss + mass unemployment, more so than AI slop or deepfakery.

[visimo-dino @VisimoD]

@NAccount32629 @Ross__Hendricks Ok, and is there evidence that people have completely stopped lending OpenAI money, and will continue to do so permanently? I’m not aware of any, but I’m open to hearing about it if it’s out there.

[visimo-dino @VisimoD]

@JohnJSSoriano @elisaxchen But endogeniety only matters if the study is claiming to show evidence of causality, which this study doesn’t, as far as I can tell? Seems like it’s merely pointing out that there’s a correlation.

[visimo-dino @VisimoD]

@flowersslop @cb_doge I’m not taking a stance on Elon, but this is pretty unfair to OP, as he didn’t say we should thank Elon for merely *being a billionaire*. Also, money and gratitude are two different things—I can’t just give my mom $100k and think that adequately shows my appreciation for her.

[visimo-dino @VisimoD]

@_JoeViben @Bunny_ngl Also, there’s a detail here that ppl seem to be overlooking, which is that the mom says “You complain every week.” So unless the mom is lying, this isn’t just an occasional thing when the child feels overwhelmed; it’s a chronic problem. Tough love is sometimes what’s needed.

[visimo-dino @VisimoD]

@_JoeViben @Bunny_ngl I disagree. Up to a certain age, sure—a parent obviously must do things to facilitate their child’s life, otherwise the child wouldn’t make it. But once that child grows up and enter adulthood, they need to learn to make it on their own, as the parent won’t always be there. (1/2)