To be fair, it did pay off in boats and vacation homes for a whole generation of salesmen.
Productivity is a collective achievement. AI in its current form increases individual output (not necessarily good output) without any way to support consensus, alignment, understanding, etc. Without AI supported collaboration, AI supported individual productivity makes everything go very fast in every direction. Like an explosion.
With the recent reverse engineering achievements, rather than zero financial payoff, it might be even be negative for the gaming sector in particular.
I don’t know.
I know Dev teams right now that have gone full AI slop programming and their PMs and Customers see it as insane productivity to the point that they have so much work coming in they can’t stop using AI now.
Do they agree the codebase is in utter shambles? Yes. Do they think it’s fixable? Yes. Do they think they have the time to fix it? Nope.
They’re all just buying time waiting for it to break. The sad part is some of them are sipping the AI koolaid so hard they say they’ll just let the AI fix it when it breaks.
I mean literally just yesterday one of the senior devs was bragging in the cube next to me that he just did 3 months of work in 5 minutes with AI.
Most of these AI sippers are old dudes who are one foot into retirement and instead of coding they want to yap about their golf club set all day.
As a younger dev and a person who loves to code it makes my heart hurt listening to the AI slop bragging.
As a programmer nearing retirement, I’m not adding to the slop problem myself, but I am absolutely looking forward to leaving the field for someone else to clean up.
I am hoping to tidy one small corner of it for an outrageous pay-out, on my way out of the field.
AI only feels like it boosts productivity because it kicks the can down the road and makes it somebody else’s job to clean it up.
You saved time generating your work email? Now I spend time translating what it’s actually supposed to say because half of it is irrelevant garbage.
You saved time generating your event poster? Now I spend time recreating it (because you’ve anchored on that version), or painstakingly editing out all the incorrect notation.
You saved time generating your logo? No there’s no way to ‘clean it up’, the anatomy and structure makes no physical sense. So I have to recreate it from scratch anyway.
You saved time automating this spreadsheet? It’s not arranged in an importable format and now i have to spend time breaking down all the arbitrarily merged cells so I can even use it. My computer freezes twice under the load of processing excess garbage data.
You only think AI saves time because it’s wasting somebody else’s.
The spreadsheet one hurts. I’m currently going over a colleague’s fancy spreadsheet and have decided that only an AI can edit it in the future, because the formulas are written in a way that need to be interacted with from a script or command-line tool and nobody’s going to do that.
Office scripts are difficult yes
There are Python tools for interacting with xlsx which are pretty nice, but I’m not going to use that to edit a spreadsheet. Kind of defeats the purpose.
Well I can give an example of why a tool like that is useful.
Export data for a client form internal system but that export looks like shit.
I could spend an hour cleaning it up (multiplied by the number of other workers needing to do similar things) or I could write an office script and create essentially an import template that formats it nicely.
I’m choosing the latter every time.
Sounds pretty reasonable actually. We could probably use something like that for some of the reports we make. I’ll look in to it.
It would require further development of AI before it can automate physical factories. There are assorted species of critters that cannot be mass-slaughtered, because they require surgical cuts to remove bones or work around tough parts. An sufficiently advanced AI can X-ray and measure the carcass, then carve up the beast with minimal waste.
Of course, this would be very expensive, requiring entirely new infrastructure and procedures to implement. I suspect that the transition to industrial-application of AI would be slow - then sudden, after testbed factories prove the viability of AI automated workflows.
oooh, nice. Universal abatoire. That’s a new one for me. Hopefully not being field tested near any schools.
Please dont give the bots factories. Thats how skynet, AM, and every other rogue AI happened.
Inside the tech industry, the company with the biggest payoff in this is NVIDIA. With another smaller payoff for AMD. That’s about it.
It’s like a gold rush for fool’s gold. The only ones making money are the guys selling shovels.
Don’t forget all the short term gains the rest of us are getting with tech debt!
Yea no shit. How could anyone have foreseen this!?
No shit it has trouble coding a simple html site how is AI replacing anyone.
Big tech isn’t exactly winning either. Over $194 billion spent with mere millions in ROI thus far.
It’s a difficult, trend, but it’s why we need everyone to get on board and use AI. If we lose millions on every billion we spend, we need to make it up in volume.
(This is - hopefully obvious - satire.)
I don’t believe this at all, the way it works is not completely direct, but through productivity gains that are harder to measure.
Productivity here is not just speed. It’s cost to get the speed.
If a manufacturer can spend a trillion dollars and get 100000% faster throughout, their consumption model may not fit that need.
Just because marketing can go 1000% faster doesn’t mean the business is 1000% more profitable given the costs
Here’s the productivity gains: https://marketoonist.com/2023/03/ai-written-ai-read.html
From the article (And OPs title)
“there are still no signs of AI boosting productivity growth.”
The AI revolution has achieved something truly unprecedented: billions in spending, endless executive keynotes, and no measurable payoff for most businesses. Apparently the killer app was convincing CEOs to transfer their budgets to seven tech companies: »Revenue per worker is rising for the Mag 7 but is flat for the S&P 493.«
I see this as a potentially GOOD thing. All those companies who dove in headfirst are gonna come out of it significantly weakened, while those wise enough to take a more measured approach will get the opportunity to take the leadership positions they probably deserve (at least more than those others). If nothing else, there might be more competition after all this.
If they didn’t lose any money to the black swan AI crash, then they don’t need a bailout.
ai has also made some people way more productive and automated some really valuable things
the entire industry hasn’t quite been one-guyed as yet. the numbers definitely don’t match the epic valuations however
ai has also made some people way more productive
People say this, but all of the data I’ve seen shows that those same people are less productive whenever their supposed productivity gains are actually measured, in any way that isn’t just “felt more productive”.
I’m open to hard data that says otherwise. But I haven’t seen it.
record numbers of “how do i connect to localhost from my friend’s laptop” and useless dashboards your devs were too busy to make, tho
Outside Nvidia. The shovel users are not making money.
