Comment by ath3nd
7 months ago
Dude, LLMs are a very sophisticated autocomplete with a few tricks down their sleeve. From that to AGI or Zuck's "superintelligence" is just light years... LLMs are a dead end for "intelligence".
We are talking about the promises of the same crop of people who brought you full Autopilot on Tesla (still waiting since 2019 when it was supposed to happen), or the Boring Tunnel, the Metaverse, and their latest products are an office suite and "study mode".
https://www.computerworld.com/article/4021949/openai-goes-fo...
https://openai.com/index/chatgpt-study-mode/
This is EXACTLY the same as NFTs, Crypto, Web3, Mars. Some gurus and thought leaders talk big talk while taking gullible investors' money and hope nobody asks them how they plan to turn in a profit.
My edgy prediction is that blunder after blunder (Metaverse, LLAMA, the enshittification of whatsapp, instagram losing heavily to tiktok and facebook having mostly dead accounts) and now giving 250M in vapor-money to youngsters, this is Meta's last stab in the dark before they finally get exposed for how irrelevant they are. Then no amount of groveling to Trump or MMA would be able to save the Zuck from being seen for what he really is: an irrelevant morally bankrupt douche (I could say the same about Elon as well) just trying out random stuff and talking big.
"LLMs are a very sophisticated autocomplete with a few tricks down their sleeve."
"Nuclear power plants are a very sophisticated steam generator with a few tricks down their sleeve."
These things are both just as true and just as informative.
Nuclear is actually a great example. It's a major scientific accomplishment but pretty much the least cost effective method of energy generation. Nuclear allowed us to generate huge amounts of energy to waste on whatever we like when the real focus should be on massively decreasing energy demands. But of course that doesn't make anyone money.
Decreasing energy demands isn't going to happen though, and energy should not be thought of as a profit generator. You can spend all your effort trying to do something that isn't going to happen without forcing people to do it, and you can lament that public needs don't make profits, or you can deal with the reality of the situation and accept that things aren't ideal.
3 replies →
Please prove that human brains don't work as a very sophisticated autocomplete.
When proving you should to prove that they are similar and not that they are dissimilar.
Let me meditate on that.
Please use an LLM and prove it yourself. Then use an LLM and prove the opposite. That's how much weight your "proofs" have.
In fact, buy a premium subscription to all the LLMs out there (yes, even LLAMA) and have them write the proof using the scientific method, and submit the papers to me via a carrier pigeon.
Other than the rockets and Starlink, I do not trust anything from Musk. Mainly because while everyone whose opinion I trust about rockets says "LGTM", everyone whose opinion I trust on all the other things he does says as per https://xkcd.com/2030/
I don't trust Zuckerberg. Bad vibes about everything Facebook ever since I went to one of their developer conferences in London a bit over a decade ago, only gotten worse since then. I wouldn't even pay for ads on their system, given the ads I see think I want dick pills and boob surgery, and are trying to get me to give up a citizenship I never had in the first place while relocating to a place I've actually left. Sometimes they're in languages I don't speak.
And while I can't say that I have any negative vibes from Altman, I've learned to trust all the people who do say he's a wrong 'un, as they were right about a few other big names before him.
And I wouldn't invest in any of the companies making LLMs, because I think the whole "no moat" argument is being shown to be plausible by way of how close behind all the open models are.
But LLMs are, despite all that, obviously useful even if you see them as only autocomplete. They're obviously useful even if you think they're just a blurry JPEG of the internet. They're obviously useful even if they're never going to meaningfully improve.
LLMs are not like NFTs. Might be like Mars, though.
Yeah... Just stay on your toes, I guess. Be open to being wrong, for the sake of you and those around you.
FWIW, "Elon Musk lies" is not a great counterpoint to an overwhelming scientific consensus.
Well, if LLMs were the next coming of Christ, where is the impact of them on open source?
Everyone claims they are awesome and super powerful in their own toy projects or in their private source. Where is the actual impact of LLMs superior power on OSS?
I can tell you:
We are drowning in slop.
https://arstechnica.com/gadgets/2025/05/open-source-project-...
I repeat again: a bunch of words strung randomly together is not thought.
Tesla self-driving aside, have you ridden in a Waymo? It’s actually pretty magical.
dude, computers are just rocks that we coerced into adding 1’s and 0’s. when are people finally going to wake up?