People are very negative about AI, and other sundries
Free post: surprising survey on AI, data problems with data, the Spanish blackout, and crisis-hunters
People are pessimistic about AI
people are very pessimistic about AI
“expert” v. normie divide, but it’s nothing like the gender divide
why exactly is everyone so negative? and where do people disagree
Other sundries
census response rates are collapsing, but is that bad?
Spanish blackout, coming to an ERCOT near you?
which investors have the quickest trigger in a crisis?
👉👉👉Reminder to sign up for the Weekly Recap only, if daily emails is too much. Find me on twitter, for more fun. People are very negative about AI, and other sundries
A whole lot happened yesterday, and Random Walk is going to write about none of it.1
This is a Friday sorta-fun-day.
People are very negative about AI
Random Walk is always a bit skeptical of survey results, but taking these at face-value, they’re pretty striking.
I had no idea people were so anxious about AI.
I mean, normie people are really bearish about AI:
35% of adults think AI will have a negative effect on the US, and another 33% think AI will be equally good and bad.
The general pessimism of regular people stands in pretty stark contrast to “experts,” who are overwhelmingly positive.
And it’s not one of those “AI will be bad for them, but good for me.”
People are even more convinced that AI is coming for them personally:
43% of US adults think AI will “harm them” personally, more than help.
For normal people, “harms” outnumber “helps” by almost 2:1. That’s just staggering, and I had no idea.
Having some concerns seems fair, but how/why did people get so pessimistic about AI?
Women are really worried about AI
Or, more specifically, how did women get so pessimistic about AI.
For all adults, 22% of men think AI will have positive impacts compared with only 11% of women. But the gap amongst the “experts” is even wider:
Only 36% of “expert” women see AI’s impact as positive
Only 30% feel more excited than concerned
At the very least, 64% of women see AI as personally helpful, but that’s still far less than the 81% of “expert” men.
What gives? Is this just male-patterned risk-loving, and female-patterned risk-aversion? Something else? It’s bizarre.2
It’s the economy, stupid
Well, it’s not that bizarre, I suppose—the concerns seem very much rooted in economic dislocation:
Just a shade over 20% of people think AI will be “very” or “somewhat” positive to how people do their jobs and “the economy” overall. That’s in contrast to the ~70% of “experts” who think AI will benefit in those regards.
I’m with the experts on this one.
I suppose I understand some apprehension about job loss, but really, only 21% see AI as “somewhat” beneficial to the economy? That’s nuts. Like, they must think all of these big companies doing AI are just out of their minds.3
Normies and experts disagree about whose job is getting axed
Another kind of interesting thing was the diversion between “experts” and regular people as to which jobs will be relatively impacted by AI:
Everyone generally agrees that:
cashiers are toast
journalists are mostly toast (lol)
engineers are 50/50
and therapists are pretty safe
Personally, I think we’ll have more “engineers” (although the role will evolve), and therapists would be toast, if the regs ever allow for it.
The people are right that lawyers aren’t going anywhere, and I’m somewhat inclined to trust their intuition around truck drivers being harder to fully automate than experts think, but it might be a bit too optimistic.
The people are probably right about teachers and doctors . . . not sure about the other categories.
Anyways, I’m not really sure what to make of this, but it’s pretty surprising. Again, some apprehension about job-loss, sure, that makes sense. Personal alienation and human connection, that too. And certainly concerns over fraud and impersonation, are obviously warranted.
But (again), only 21% see AI as “somewhat (or more)” helpful to the economy overall? That’s bananas.4
We haven’t lost the thread. There is no thread.
Other sundries
Just a few other unrelated observations before we go.
People are worried about response rates
Random Walk has observed declining response rates to BLS/Census survey work before, but Odd Lots did a whole podcast on it and it’s good.
“Some of America’s Most Important Economic Data is Decaying.”
Gathering official economic data is a huge process in the best of times. But a bunch of different things have now combined to make that process even harder. People aren't responding to surveys like they used to . . . Bill Beach was commissioner of labor statistics and head of the US Bureau of Labor Statistics during Trump's first term and also during President Biden's. On this episode, we talk to him about the importance of official data and why the rails for economic data are deteriorating so quickly.
Survey response rates are, in fact, cratering:
Response rates to the CES survey dropped by almost a third, from ~60% to ~40%, and is yet to recover.
Interestingly, lower response rates are not unique to the US, but “also plagues other advanced economies (Flodberg and Wasén 2024).”
Even more interestingly, the SF Fed concludes (and I’m paraphrasing): response rates tanked, but it hasn’t mattered much.
Just by looking at the revisions to the headline statistics, these have all been in-line prior averages:
Because the response rates of many household and business surveys have been declining, the incoming data may have become more uncertain and less reliable than in the past.
Our Letter shows that this is not the case for important labor market and inflation measures. Over the past two to three years, the revisions to payroll numbers and CPI inflation rates have been in line with their pre-pandemic averages.
No harm, no foul.
Truthfully, Census data is getting less useful relative to far more timely, and relatively higher fidelity, “alt” data, e.g. credit card data, location data, job posting, etc.
Spanish Blackout coming to an ERCOT near you?
Spain had a big blackout because something fried the grid.
I have neither the time nor the inclination (nor the expertise) to get to the bottom of this one, but some people are speculating (with variously high degrees of conviction), that “going green” has a lot to do with it.
Could be, maybe.
Certainly, Random Walk has previously observed that more knowledgeable people have observed that solar and wind are both intermittent, and that that intermittency creates costs, challenges, and complexity for the grid, but transmission specifically.
So does does going green make you blackout? Idk.
But that’s just a segue for this little doomcast from Factset on the coming shortages to ERCOT (the Texas grid):
Loads are expected to exceed capacity for ERCOT sometime in early 2027.
That gloomy forecast even contemplates that all existing renewable projects are completed. So, it’s either a lot more nat gas, or blackouts, I guess.
When panic strikes, there’s one investor with a finger on the trigger
One last thing that’s in the spirit of coming doom, but also somewhat amusing (imo).
The question is: what happens when crisis strikes? Do investors call capital from their LPs, or not so much?
Well, looking at the pandemic as a prior, the answer is “kinda,” but one investor really stands out:
When the pandemic hit, distressed debt funds called over 20% of uncalled capital (which is ~8x what they normally do on any given month).
Those Distressed bois were ready. Hot damn.
Truthfully, all the asset types were a little trigger happy (which makes sense), although presumably those were defensive measures, for the most part (as opposed to opportunistic).
But hey, there’s no more important time to have rescue capital than during an apocalypse, so if an apocalypse is your time to shine, then shine bright, bight like a diamond.
Stay sharp, Distressed bois, your time may come:
Polymarket is now giving a 66% chance to a 2025 recession.
What do those degenerate gamblers know, anyway?
Other reads
Laid-off federal workers hoped to land state and local government jobs. The reality is messy. Because the process is cumbersome and bureaucratic, oh irony of ironies.
China doesn’t want America to see its Trade-War pain. No, but it’s definitely got some.
British asset manager has ~$10B outflows, as clients reduce exposure to China. More evidence that tariffs are “working.” Again, China needs the world as buyers just as much (if not more) as the world needs China as a seller.
Ares raises $3B for semi-liquid PE-stakes fund. I would expect to see more of this. It opportunistically takes advantage of the liquidity demands from existing PE LPs, while alleviating illiquidity concerns from new non-institutional investors. I suspect the fees are probably pretty high.
Duke offers buyouts in cost-cutting push. Trump admin’s initiative to create accountability for higher ed is bearing fruit. Good to see.
Apollo raises $5.4B debut secondaries fund. Because all the action is on the private side.
Previously, on Random Walk
Volatility begets volatility, and other sundries
Vol begets vol (and the analysts get it wrong, a lot)
Healthcare makes all the jobs . . . but maybe manufacturing too? Other labor market desiderata
healthcare jobs replaced all the manufacturing jobs (much to the benefit of women and aides)
Random Walk is an idea company dedicated to the discovery of idea alpha. Find differentiated data, perspectives and people, and keep your information mix lively. A foolish consistency is the hobgoblin of small minds. Fight the Great Idea Stagnation. Join Random Walk. Follow me on twitter. Follow me on substack:
McDonalds missed on the continuing decline of same store sales (which were worse than they’ve been in a while). That’s definitely a ‘real economy’ signal to worry about (although transaction data is still strong). Reddit and Instacart both did great. Amazon did ok. Apple did great, but the street thinks it’s front-running. Square/Block did poorly. KKR lost money because it’s fixed income lost money on higher yields. Airbnb did great, but gave a soft guide.
Among experts, men and women agree on basically two things: the risk of impersonation and the risk of job loss.
Everything else worries women far more than men.
The people are also really bearish on personal relations, where only 7% see any benefits. I suppose I worry about that too.
Random Walk is generally hesitant to “blame the media” for negative or positive views that are untethered from reality. It’s a cope that gives people far too little credit, and the media far too much credit.
Where I suspect “the media” matters most, however, is on subjects where the “truth” is furthest at-hand. That’s because it’s easier to have to “false” beliefs when consequences are relatively attenuated, and vice versa.
Is that applicable to AI? Idk. Media certainly can help spread paranoia and negativity, tht’s for sure. But while I’m pretty hard on the media (bc the media deserves it), I wouldn’t necessarily have pegged them as anti-AI, but maybe I focus too much on markets-related stuff? Maybe it’s some other part of the information mix? But also, LLMs are pretty broadly available to play around with, so it’s not like firsthand experience with them is so hard to obtain.
I know the details about the Spain blackout. It wasn't intermittentcy, it was "contingency plan implementation coordination failure". It was similar in multiple ways to what happened in California with water during the recent LA fire catastrophe. That is, the people responsible for planning for how to deal with certain crisis scenarios identified the need for certain minimal amounts of critical response resources, but didn't take the extra step of making sure the people with control over those resources would know to never let availability fall below that threshold. Whoops! Let me explain.
In general it's not intellectually productive for green-policy critics to reflexively blame "renewables" when an electrical power supply system is required by policy to use lots of them, as it can still be engineered to be robust to all the special challenges posed by those sources. The only real question is price and whether policy makers are going to accept those much higher prices and be honest about them, or else try to keep the prices down by reducing how robust it is, that is, saving money by taking on extra risk.
So, for renewables there are two major problems that can be engineered around. Intermittentcy can be dealt with by being able to suddenly import lots of extra power from elsewhere (Iberian system could not) or crazy amounts of batteries (too expensive for most situations) or having plenty of gas reserves and idle gas turbines that can be quickly spun up to fill the gaps. And the Spanish system seems to have been built to be up to that challenge.
The other problem is frequency. I'll lose some accuracy and precision to spare you the technical details, but basically, AC power in a system runs at a particular number of hertz (50 in Spain), and everything is built to deal with that exact frequency, and lots of very important and very expensive electrical stuff everywhere will fry and start fires if the frequency gets even a little bit off target, and so they will just disconnect from a system if that happens, and can't use, transit, or contribute power until the frequency is restored and stabilized within a fairly tight range.
In a big failure, the frequency drops to zero and needs to quickly come back up. The problem with most solar installations (and sometimes wind, it's more complicated) is that the solar power can't help much with this, indeed, the solar power has to sit on the sidelines and 'wait' for system frequency stability until it can join the party. What does the job is by using the angular momentum inertia in giant heavy spinning turbines and generators in either huge fossil fuel sites or hydroelectric dams, which you can imagine being like a giant electrical flywheel that keeps everything going with inherent frequency stability that resists change. If it's big enough, it bootstraps the whole system back into shape in a second.
Spain no longer has enough fossil plants to provide this inertia everywhere it might be needed, but that's OK, thought the engineers, because they've got five big hydroelectric dams, and those can provide enough frequency inertia to deal with any grid crisis.
Unless a bunch of uncoordinated decision makers had taken three dams offline for maintenance at the same time. Dios mio! Ay caramba!
The remaining two weren't enough to get things minimally going again for about another 10 hours, and by then, it's dark, so no solar, which increased the delay to getting back to fully operational status.
Moral - Planers need to think about the meta-contingency of how their contingency resource availability might fall below threshold because of uncoordinated human decisions if control over availability is decentralized. California Firefighting policy and Spain Electrical policy failed to provide for this.
"The people are probably right about teachers and doctors . . . not sure about the other categories."
I think a lot of doctors are in trouble unless the guild structure manages to close off AI supervision. Proceduralists like surgeons and such are probably fine for the near to mid term (though I suppose automated robotic surgery may be coming down the pike soon). But I can really foresee nurse practitioners and physician assistants working in clinics and providing care in hospitals by doing the exams, talking to the patients while being supervised by AI. Perhaps there will need to be a few supervising physicians still around to supervise the AI initially, but a lot of that care is supposed to be evidence based and to conform to algorithms anyway.