MIT says AI isn't replacing you it's just wasting your boss's money
interviewquery.com[dupe]
Another day, another substa...
Lots of discussion just a few weeks ago
https://news.ycombinator.com/item?id=44941118
Thanks! Big discussions too.
95% of Companies See 'Zero Return' on $30B Generative AI Spend - https://news.ycombinator.com/item?id=44974104 - Aug 2025 (415 comments)
95% of generative AI pilots at companies are failing – MIT report - https://news.ycombinator.com/item?id=44941118 - Aug 2025 (167 comments)
AI is predominantly replacing outsourced, offshore workers - https://news.ycombinator.com/item?id=44940944 - Aug 2025 (235 comments)
The actual report mentioned in the link
https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Bus...
Just waste it on me, boss. In the long run, I’m cheaper
Cynically, it doesn't matter if AI actually replaces jobs, but as long as the stock market BELIEVES that it does, then it gives companies permission to do layoffs without taking a hit to the stock price.
I don't see anything strange going on, market just finding out how to properly milk the new tech. Where it is useful and where it is not.
Another blog post about the study that found 90% of employees were using AI regularly in their work....that totally omits the part about 90% of employees using AI regularly.
The study found that company implementations were failing while people heavily used their own choice LLMs.
I agree. The original report is vague about that as well:
> Over 80 percent of organizations have explored or piloted them, and nearly 40 percent report deployment. But these tools primarily enhance individual productivity, not P&L performance.
How on earth does "enhancing individual productivity" not improve P&L performance?
Snark aside, this is the main topic of Eli Goldratt's "The Goal" - one of my favourite books. Output of the business is constrained by some bottleneck. Improving efficiency away from the bottleneck is just waste. Said differently, enhancing individual productivity has no effect on the business output unless doing so elevates the bottleneck. Sadly, most business don't know what their constraints are or are otherwise blocked from elevating them due to politics and other factors.
You have to factor in the cost of the tools.
And also, just because an employee is producing more code or more reports or more trades or more summaries per day, that doesn't guarantee the company will make more money. If you enabled your market analysts to make twice as many trades per day, you will likely lose money if they actually did that. Because they were already finding the good trades, and now they are also making the not-so-good trades and maybe even the unlikely trades.
And also, in the longer term not being shown in existing studies, you are in an arms race with your competitors. Like advertising, you will be spending a lot of that money in order to counter the spending of your competitors, and you both lose the game of prisoners dilemma, the only winner being the arms dealer selling to both sides.
How is this 'enhancing individual productivity' being measured? If by self-reporting, it's worthless.
Here's a recent experiment: https://www.theregister.com/2025/09/04/m365_copilot_uk_gover...
A fairly common pattern seems to be emerging; users _believe_ that it makes them more productive, but attempts to actually measure this don't show that.
> How on earth does "enhancing individual productivity" not improve P&L performance?
If the individual is more productive, they can work fewer hours. This is especially true in a remote-work environment where you are not in an office setting.
>If the individual is more productive, they can work fewer hours.
I have never seen people working fewer hours due to technological improvement, productivity just goes up from them working more efficiently for 8 hours.
Have you met many humans? ;)
A very large proportion of the people I've worked with over my career would absolutely work less hard if they could get away with it.
Ah for sure, that's not what I meant. I was talking about the fact that sooner or later people running the business figure out they can increase the workload if employees have too much spare time. Else we'd be at pre-computers productivity levels.
Fair point!
If your boss doesn’t get credited for it it doesn’t count?
That was a random show thought I had a couple days ago:
> We’ve managed to convince the biggest tech companies to invest trillions into tech that doesn’t replace SWEs, rather, a new dev tool at our disposal to make some aspects of our job easier: search and tedious coding tasks, saving us time.
> saving us time
citation needed
where my boss's money comes from?
Not this again
If you are not already viewing work as a transfer of wealth and nothing more, nows your time.