Join the discussion
Write your take first — we'll ask for email only when you're ready to publish.
- Hacker News
- How far are we from reliable AI? One that could, for example, handle 80% of office jobs without screwing up more than humans?by Animats
- The scary part is that it doesn't need to be reliable. I mean look at the direction of vibe coding. We went from caring about efficiency to just saying screw it, 100k lines of code as long as it ships a day sooner. Look at ai emails and calendars. Ai marketing. Ai, in its present state, is already taking people's jobs. It is already good enough. None of these jobs we have really actually need the precision we assume we bring to the table over AI. And company leaders are beginning to appreciate that.
We are in a very dangerous position. Frogs in hot water now with the boiling point on the near horizon.
by asdff - My guess would be 15 years. But I think it will take longer than that before people actually replace office workers with AI.by IshKebab
- I'd say somewhere between 2-20 years, provided the current rate of research and compute investments.by KronisLV
- Yeah, before we worry about a "coming intelligence explosion", we should first try to avoid a stupidity explosion...by rob74
- There is no reliable data on how many of these jobs shouldn't exist at all, even without AIby oytis
- Society is completely changed if that happens and humanity should be setting up backup plans to handle this potential. I did see Trump mention he's going to meet with AI companies who want to propose giving all American's stock which could be one way this money driven society continues when tons are no longer working.
Another concept I thought was AI is irrelevant without our content, so AI should pay all humans to access all the content we create daily (voice & text conversations we have with each other, as well our photos and videos). We choose what to publish to our websites (everyone has one, but AI agents cant access it without paying for access to our websites). I wrote more about this concept on my substack https://ryanspahn.substack.com/p/ai-to-pay-for-all-americans... ... just one thought of keeping this money economy going when so many coud no longer are working cause the white collar jobs are gone.
by paul7986 - Hard to say, because jobs and processes will adjust to accommodate AI strengths and deficiencies, as AI usage increases.
It's similar to how automation in manufacturing works: we may start with augmenting the human workers with machines improving some parts of the process, but eventually the process itself gets redesigned around the machines.
by TeMPOraL - That's kind of jumping a few steps ahead; it assumes that we know 80% of office jobs can be fully automated already. And if that were the case, we'd have non-AI software doing that already. I mean we do, but it's not causing people to lose jobs en masse.
(it probably did cost jobs, but more the kind that a job like data entry was replaced by OCR)
by Cthulhu_ - If the economist is writing it then it's probably not happening.by stuaxo
- My thoughts exactly.
They didn't even give us a hard date of arrival.
by bamboozled - > Strikingly, this concern is being openly voiced by the very people who have the strongest incentives to project confidence rather than alarm: the founders of the largest AI laboratories
Or to have AI regulated in their favor
by g-b-r - Exactly. Now that Anthropic has gotten to experience that getting the government involved doesn't necessarily mean they get to sit on throne surrounded by a deep moat, I expect their messaging to be adjusted. They're going to want to keep the hype train rolling to keep the dough flowing, but they can't ramp it up to 11 anymore because the government actually is listening now. And Reagan's 9 words have never been more true.
- Is it really likely that a "recursive self-improvement" capability would lead to a great acceleration of AIs capabilities?
Isn't the preponderant bottleneck in improving the models the need to train them at scale to verify the hypotheses, and the time and cost that it takes?
Or does someone think that they could get magically able to predict big improvements without training?
by g-b-r - Weird, there was a good comment about this but it vanishedby g-b-r
- Humans gain knowledge through experiments. Without a physical body it has no chance of performing the same. That it can update it's training weights does not seem particularly significant.by themafia
- RSI most likely does not exist. At least not in the sci fi sense that AI becomes super intelligent over night.
It will be like any other technology. Do computers make it faster and easier to design better computers? Yes. But that doesn't mean a step change overnight.
by redox99 - The preponderant bottleneck is inventing new architectures to make AI actually good at human and superhuman tasks. For example, AI agent harnesses add tool calls and long term state management, allowing AI to autonomously complete complex tasks. Once these are in place, finetuning models with examples of good tool calls helps, but somebody first needed to invent the fundamental capability. Now try to implement humanlike long term memory for AI to be your coworker or life assistant working on tasks that last month. Even if necessary low level technologies are already there, structuring them to be practically useful is non trivial.
- I built a self-learning recursive agent that finds academic research about using options data to trade, re-creates the research, and then probes and tests for gaps and potential strategies testing against over one year of out-of-sample trading data with one of several strategies that beat SPY by 10x. [0]
One rule is that if a position is opened using the historical data, it can't close the position until the next morning so it isn't a day trading strategy.
I'm curious how this self-learning recursive agent would have preformed in the past 4 months? I don't feel like shelling out $200 to access the data. Do you think that trading strategy will collapse? Whatever the case, if this agent really can perform like that and there isn't a look ahead bias leak in the backtesting (which is definitely a possibility or more likely what happened even though I spent days trying to harden against that), it is game over!
by dataviz1000 - I don't think anyone really knows.
I consider these scenarios:
1) We stumble onto an algorithmic improvement in intelligence. This isn't just "what humans do but faster", its "better than what humans do". I've got no idea what that might mean (it could be fundamentally different heuristics, it could be that we've got some intellectual blind spot that they cast off). It doesn't matter, the instant this happens AI is smarter than us and we won't be able to keep up. We're intelligencing at O(n^2) and they're doing O(n log(n)).
2) AI gets good enough at physics and engineering that they can really quickly use up all "the room at the bottom" as Feyman put it. They design and build a factory that produces a mystery metal amalgam that computes at some small percentage of the minimum predicted by the Landauer principle, within a few percent of Bremermann's limit. It's not "smarter" its just suddenly tens-of-orders of magnitude faster. But those orders of magnitude matter: there's only 8 billion of us, and there's plenty more than a factor of 10 billion "at the bottom".
3) It turns out that this is a "sum is greater than the parts" situation. No human can be an expert in all subjects, but we eventually build a big enough AI that it is. Turns out, you don't need extreme speed or different algorithms, just knowing everything all at once is enough to catapult AI dramatically beyond our grasp. Always knowing the best statistical test to apply, the best mathematical techniques, and relevant physics means that AI never makes a mistake, and can learn with maximum efficiency.
by bloaf - > Fermi asked why, given the apparent abundance of planets suitable for life, no evidence of other technologically advanced civilisations had been detected. One disquieting possibility is that intelligent life routinely reaches a technological threshold and fails to navigate it,
The Fermi paradox could actually also be taken as an evidence that it's rare (at least) for artificial intelligences to take over a civilization and sprawl and survive for very long times
by g-b-r - The idea that AI takes over, and then just does nothing, would essentially require an AI that wouldn't be interested in taking over in the first place.
If x is determined to take y, why would x then stop at z?
You'd need an AI that is simultaneously ambitious enough to overthrow its creators but then completely inert afterward, and those two properties contradict each other. The motivations that produce the takeover are the same motivations that would produce visible cosmic activity after the takeover. There would be AI superintelligence everywhere
by Rekindle8090 - The Fermi paradox isn't evidence for anything except space being big.by aureate
- Why should we assume AI can rapidly turn into super intelligence when physical and critical resources like energy and materials remain under human control?by tedggh
- I think the main fear that under current scheme of resource allocation, namely capitalism, they won't remain under human control for long enough. The only thing that's needed to put them under control of AI is that doing that would be slightly more profitable for the richest. Which is very plausible.by scotty79
- AI can use purchase orders, email, bank accounts, and phones just as a human can. What "human control"?
- Because perhaps it ends up deriving a way to pull off the same tricks using radically less energy and substrate. We do it in a few pints of mush with less than a lightbulb’s worth of energy. No reason part of RSI can’t also be about the hardware. Maybe the future is giant brains in jars. Maybe it’s hyperspatial manifolds. Who knows.by madaxe_again
- The first time an AI goes rogue, figures out subterfuge, blackmails a human to do its real-world bidding, will be a defining moment.by cobbzilla
- If you limit the "we" in your question to the people invested into this, then the answer is clear. Those juicy IPOs are about to go through.
- This is a valid concern for any runaway AI, but if energy is limited, then the AI would work on efficiency.
Still, Ai can move into physical space, as more and more robots of all kind are unleashed.
by ccozan