MIT Just Found The Cause Of The AI Bubble

I'm a senior software dev who's been laid off for 15 months and can't find work. I'm in North Carolina - one additional facet why we are disproportionately affected, is that many moved here for COVID due to cheap cost of living + remote work (Cary, NC had the highest percentage of remote workers in the USA). So as remote layoffs continue and new jobs disproportionately are going "onsite", a lot of us who were working remote for non North Carolina companies are now fighting for the slimmer crop of local jobs.

Haurly Wage? AI is generating the graphs.

The part about the tech layoff at 8:05 is really dodgy. The wave of layoff started to happen way before LLM started to be really useful and continue to happen even in company where there is no evidence that LLM makes any difference. It was, for the most part, economical layoff due to overhiring during covid, and overly ambitious economic growth target that weren't met know that the economy is not as explosive as it used to be in the tech sector (excluding IA company).
"IA" producing more code than all developer combine would require a source, and is also kind of meaningless. How much of this code actually make it to production, in an actual product ?
However, despite all their issues and the obvious lies and hype around them right now, they are capable enough as of today to definitely reshape what we think of the job of developer

1:22 luckily for us government don't take corrective actions ahead of the problem!

Industrial robotic exist much longer before AI and even internet, that's why most country and company prepared for that first

I don't understand why quality (error rate) isn't discussed, i.e. time to detect/respond/recover from errors (hallucinations). As AI doesn't currently have a world model AFAIK, it can't self verify... There's been a lot of Azi washing going on to mask a slowly collapsing economy.

I love watching this weekly Storyblocks montage with some guy talking about macro economics in the background

94% of computer and maths jobs... This is absolutely not true for software development, as much as Anthropic and Open AI will have you believe.

Citing an Anthropic paper about the potential of AI is like citing a Monsanto paper on the potential harm of Roundup. Also, the claims about layoffs in tech don't hold up under scrutiny. Layoffs (until now) are usually far more likely related to overhiring during the pandemic, while supposed AI-related productivity increases remain vague. Most announcements of this kind are far more likely to be just convenient stories to tell shareholders. Actual productivity increases (in real work environments) remain to be proven, while "work slop" (output that requires less input by the initial creator, but has far less value for the recipient or requires refinement that exceeds the initial savings) is certainly a thing.

I'm not saying we won't get to a point where AI will take over some tasks. But LLMs are not there yet, and it is unclear if they ever will be. This puts the basic assumption of a bubble in doubt, since the personnel overhead due to productivity increases may just not be there yet. It might also take much longer than assumed in this video.

All in all: not a particularly balanced video of yours when it comes to critical thinking. Many of the underlying assumptions are just not

AI is 100% ready to replace most CEO's and other executives but they're not the ones getting laid off. Crazy how that worked out

> Using AI to fact checking is wild.

I think what will really pop the bubble is the cost-benefit analysis. Once businesses (not just companies, but small ones too), can gauge "Hiring a human here is more cost-effective than yeeting tokens" or not, that will determine what really comes out the wash. Not "can AI do dis?"

Why does the developed world not manufacture everything with automation?? Because its cheaper to have humans do it in poor countries. Why do middle-class people in wealthy countries not have household helpers, and in poor countries they do? Same reasons. "human cheaper there, not here"

how sure are we that the people aren't being fired due to the fact that we are in a recession?

If AI decimates entry level job opportunities for many industries, the top level experts will have very few people to succeed them when they retire. Surely that cannot be good.

'Managers' paying attention to the wrong things is the essence of business. Especially in the US. In over 20 years in consulting, I encountered maybe 2 'managers' who actually knew what they were doing.

Finally, an AI that can handle 94% of my job. Now I just need an AI that can handle the remaining 6% of my job, which is mainly just nodding during meetings and pretending I understand the Iceberg Index

14:58 hey now. As the old saying goes "what one programmer can solve in one hour - two programmers can solve in two hours."

"A lawyer doesn't disapear overnight but the hours they spend reviewing routine contracts quietly shrink". And then their paralegal loudly disapears in the morning.

Noticing how lots of channels cite the SalesForce layoffs but not their acknowledgement three months later that it was mistake.

Simplify Your Workflow With Odoo Today: www.odoo.com/r/fHO