I don’t know anyone in the tech industry who thinks AGI will never happen or that software engineering and white collar jobs can’t be automated. We all read sci fi, you’re not some unique visionary for anticipating AI. The frustration is with how much the claims have outpaced reality and how poorly the investors and executives have treated their workers during this transition.
My (probably flawed, but still) memory is that the first one of these threads I participated in, at the end of 2022, was saying that none of us would have jobs after two more years of development of these models. Two years from then was almost two years ago now, and we're still "a year or two" out.
On the flip side, the thesis that these will be useful tools that will augment the work of software developers when understood and used for the things they are good at has (IMO) remained undefeated during this entire period.
He's not claiming there's no flaws or that there aren't CEOs claiming more than is possible right now. He's making the point that these don't negate the parts that are genuinely impressive.
I've encountered waaaay too many "you can't possibly have done that with AI because AI occasionally hallucinates" and "CEOs say things for marketing therefore AI can't really do anything" type of posts.
Everyone is wondering about what happens if (when?) it finally comes true
We really have to figure out what comes next for your average person. I think the reason no one wants to talk about this is because the answers aren't great. The average person not going to be living a great life in all likelihood, once we have no access to capital anymore
This still backfire to the oligarchs
It's probably a really bad idea to keep building them autonomous killing machines though
Albeit that is because I did a BSc in psychology where I developed a deep distrust for intelligence research and concluded that intelligence is not a useful term in philosophy nor science (and especially not in engineering).
We are so much closer to 1G constant acceleration space flight than we are to AGI. We know, in principle, how to achieve 1G travel. We don't know, in principle how to achieve AGI. Our best guess so far is something along the lines of "emergence" which means "maybe if we do enough matrix math in the right way it'll wake up and become a being with agency and intelligence". Another way to say this is "hopes, prayers, and lots of GPUs".
Let's all get a grip. Without a coherent theory of intelligence, you aren't gonna make one in a lab. That's not how science works, it's not how engineering works. Start at the beginning.
[0] https://en.wikipedia.org/wiki/Project_Orion_%28nuclear_propu...
I mean, we did it and there have been roughly 117 billion human beings with GI in all of existence.
Like if you feel it's not important enough to write yourself, just don't put anything there?
They have genuinely been conditioned out of being able to write a sentence without the help of their thinking machines.
We haven’t moved past this yet
If AI really is at human level quality/error rate (I don't think it is for general tasks, but there are some areas where it is), then the answer is typically cost and speed/capacity.
Remember, we aren’t just talking about the product you create. While you would measure deliverables by cost and speed are we ignoring something else? Something that could potentially be more important than either of those metrics?
Yes. How long it'll take and how much it'll cost are going to be among pretty much any customer's first questions.
They're not the only considerations, and could potentially be outweighed by other concerns even when quality is the same, but I think they are the main drives of AI adoption in industry. If error rate is the same, a $1/hr (amortized) camera and machine vision model capable of checking 300ft of material for defects per minute will likely be preferred to a $10/hr human QA capable of checking 30ft per minute, for instance.
But that’s not the argument here, is it?
So the question still stands.
> > People try to excuse AI failure modes by saying humans have them too, but if they're the same then what's the whole point of replacing a human worker with AI?
To which my response is that speed and cost are also important factors, which can often give AI the edge in considerations when quality/error rate is equal.
If you meant something other than that, you may have to specify.
See: traffic lights
If I have a choice between a deterministic traffic light and a non-deterministic traffic light which one would I use?
And yes, before you say “this isn’t a comparison of non deterministic and deterministic tools, this is a comparison of two non-deterministic tools” think about what my next question might be.
Boston consulting group Bain and MacKenzie make billions of years completely making shit up. same thing with Ernst and young and any of these organizations that make these “future of (insert market)” reports
I'm in the pagerduty lineup, and shockingly, my IQ isn't even a mere 185.
It's like human to dog years ;-)
Right now single prompt with Fable can get us a small protype of like 1 game mechanic that's not even remotely production ready. So this guy thinks it'll 100,000X in a year.
AI boosters are something else.
But maybe my particular skill set where all those roles were only really out of reach for me for time-constraint reasons is less common than I think… I dunno, though, between people who’d moved up into managing projects but can drive an LLM pretty competently for programming (ex programmers) and versatile can-talk-to-people seasoned programmers, it’s all the other roles that look increasingly like they’re slowing productivity, rather than increasing it, now.
Someone on a call the other day tentatively brought up that they were noticing it was taking longer to get all the paperwork right for a bug fix or even mid-sized feature than it was to actually write and test the fix, by the time they looped in some variety of person in a jira-wrangler role. It’s clear those jira-wranglers are gonna have to fight to keep their jobs (I don’t really know how they’re gonna do it, I feel apprehensive on their behalf in every meeting now)
It’s just that now if I have an LLM that can talk to my business comms and reporting tools (jira was already mostly for managers to generate reports, not to help the people doing the work, that’s why it sucks so much as a tool for getting actual work done) and my first-pass at programming looks a lot like just writing a programmerly-flavored feature or bug ticket, having other people making useless first drafts of tickets for me to rewrite (and probably have to go clarify them with an SME or stakeholder anyway) and someone dedicated to poking around in Jira in general, is gonna start looking funny even to bigcos, I think.
I assume that’s entirely due to not being able to downvote submissions on HN.
As for the article, as another user put it:
> Call me when it stops making things up.
We haven’t moved past this yet
He's not my boss, so he can't literally tell me what to do. And my actual boss has told me to ignore him. But it's a worrying but of psychosis that I fear could infect the rest of the C-suite of it isn't addressed now.
The bad news is it’s already in the C Suite and they (were) pushing it hard.
The good news is that since Copilot and others have started to charge based on usage a lot of those same leaders have hit the brakes and are now wanting detailed information and usage reporting to figure out where AI actually fits in.
Some have gone even further and slapped a usage limit on individuals or teams and left it at that.
Sanity is around the corner. At least until the next big thing :)
Passing specific tests to the point that the internet is now full of "Is that content AI generated or not" debates? Yes
We currently have debates around whether content online is human or ai generated and instead of acknowledging the milesetone we have a post essentially claiming "the turing test was passed long ago, it's worthless".
1. this is prone to taking the fringe perspectives and making them "everyone" 2. most of the points are highly highly interpretive (growth could be pressure, laziness, FOMO, self deception, real valid growth.. we don't know yet)
LLMs are definitely the most transformative thing I have seen in 25 years in tech, but I still think like every other hype cycle there is a lot of lying and self delusion, I don't really think any of us, neither myself nor the author really know what we have here.
We’re treating LLMs like we have every other tool. It’s going to become like spellcheck eventually, not something you think about, and certainly not worth hundreds of billions.
It’s the economics that make no sense. That situation has been demonstrably getting progressively worse.
> An AI notices an unmet need, builds the product, finds the customers, and runs the company to a billion-dollar valuation with zero employees.
I'm OK with this. Delamain is awesome.
I’m gonna send this to people when they tell me that nothing‘s happening and none of this is real
I love this so much thank you
AI is already better than us at a bunch of things, and worse at a few. The list of things it's better than us at increases every month.
In 5 years, people will still be saying "Well, I can still ride a unicycle blindfolded better than a robot so it's not AGI."
AGI is such a meaningless term and puts too much importance on human-level intelligence.
My point was that there's nothing objectively special about our level of intelligence, so it shouldn't be used as a benchmark.
That sounds like how tech CEO's treat their employees!
Kidding aside, I think the notion that you could control something (legitimately) smarter than you is a pretty risky proposition. (Fortunately one that I think is actually far off)
But then think about the CEOs and their employees again.
We just need to invent something like money for an "AI" and then we can lead in on a stick ;-)
While I think the discussions around alignment sound more like science fiction than science, it also hasn't been "solved", and the only "control" you have over what the agent does is by gating access. Even if you think the AI itself is aligned, handing complete control of your entire enterprise and all your data to a private unaccountable for-profit enterprise should give you a lot of pause.
The idea that you can create entire jobs where humans just act as reviewers or gate keepers seems very unlikely to me, just knowing human psychology. If the AI is right 99% of the time, and catastrophically wrong %1 of the time, but I've been condition to always hit "accept" because it's usually right, there's very little chance I'm going to catch that 1%. "yes, continue with the database migration". "yes, continue with the database migration." x 1000. (At button click 500, the user stopped reading). "yes, go ahead and delete the database. Shit, wait!!!"
Well, Trump seems to control a lot of people given how afraid they are.
Trump is also called not the smartest person out there.
So...
Can you though? I've seen some videos recently of some pretty darn scary Chinese robots that I suspect might already be capable of riding a unicycle blindfolded if someone set 'em on that task. ;)
1. At least humans can still do highly specialized physical work
2. At least humans are cheap to run on low-value hard labor tasks
3. At least humans are desperate and highly expendable, in some types of work it's cheaper to 'use one up' and replace them with another than to run and constantly repair robots
4. Humans are useless for all work and therefore have no value
In software, it hasn't replaced all the software engineers (even though it's better and faster than us), it's just meant that we now have more leverage.
We can write more code and work on more projects than ever. I don't see why we would suddenly stop using it like that?