Questioned

I, for my part, am not a fan of populist politics, mainly because everyone can be at fault but "the people." Granted, this can said to be true of representative forms of government more broadly, but populism tends to explicitly make it front and center, as candidates running on populist platforms seek to burnish their "the people can do no wrong" bona fides.

Enter James Talarico.

“Why do we have a trillionaire when veterans are sleeping on the streets? The reason poverty exists is not because we can’t feed the poor, it’s because we can’t satisfy the rich.” By the way, what happened to the point of the question mark?

Personally, I would have figured that he's smarter than this, but, to borrow from Upton Sinclair, "It is difficult to get a man to understand something when his political ambitions depends upon parroting the public's misunderstanding of it."

Big picture, the "reason 'we' have a (paper) trillionaire" is simple. Given the choice between spending money to buy goods and services from one of Elon Musk's companies and spending money on subsidizing housing for impoverished veterans, more people choose the former. Enough people choose the former that the value of Mr. Musk's interest in his companies is astronomical, because the Boards of Directors of those companies believe (correctly or not) that a) no-one else can increase the value of these companies like Mr. Musk will and b) to keep the man's mind on the task at hand requires showering him with oodles of equity.

That's it. It's all down to choices that people have made. And if the United States, as a society, made different choices, there would be different outcomes.

No-one is prying some homeless person's last dollar from their desperate grasp to hand it over to Mr. Musk. People don't feel that they're well-off enough to devote resources to the issue of homelessness and/or they don't feel that the money will be well-spent/generate workable returns. In other words, people with resources are pursuing value, rather than charity.

I get that Mr. Talarico is playing on the sense of poverty that Mr. Musk's current net worth is generating, but this way of putting things gets in the way of people understanding that they have at least some agency in fixing it.
 

Traded Away

(Note: There are some long quotes here from the original story, as I'd like to preserve the context for my analysis.)

So, to make a long story short, a now 35 year-old man who's only identified by a middle name, Thomas, was a problem gambler racking up tens of thousands of dollars in debts. After bankruptcy, he had himself added to his state's self-exclusion list, so he couldn't sign up with sports books again. NPR relates:

"Then one day, about two years later, I'm scrolling on Instagram and I came across an ad for Kalshi," said Thomas. "I was bored and thought this could be fun, and it ended up turning into something that wasn't fun at all."

You can guess what happened after that... more tens of thousands of debts piled up. Realizing that things were getting out of control...

"I asked a live agent to close my account. I have a gambling addiction and have been self-excluded from regulated gambling platforms for years," Thomas wrote the company in a chat feature of the app, according to a log of the conversation shared with NPR.

An automated response from Kalshi noted that "self-exclusion is a responsible trading safeguard designed to be irreversible." It offered other solutions: "Three different responsible risk-management actions can be taken: trading break, voluntary opt-out, and a personalized funding cap."

Thomas insisted he wanted to be banned from the app. Two days later, he also made three separate requests to Kalshi over email for his account to be permanently closed, correspondence he shared with NPR show.

Kalshi did eventually bar Thomas from betting.

NPR went to Kalshi for an explanation.

Dani Lever, a Kalshi spokeswoman, called Thomas "a cherry-picked case."

She went on: "If you ask our millions of traders, they would emphatically tell you that an exchange model is significantly healthier than a sportsbook model: our profits aren't tied to trader losses, so we don't have the same predatory incentives."

At this point, I found myself wondering, if Ms. Lever claimed that Kalshi didn't have "predatory incentives" to keep Thomas on the platform, why not simply let him sign up for the self-exclusion list? If Kalshi didn't need his business, why the hoops to jump through?

"We've prioritized making Kalshi the safest venue for people to trade on," Kalshi's Lever said in a statement, pointing to features that nudge bettors to take a break, establish limits on what traders can deposit and partnerships with mental health counseling to assist problem traders.

But again, why not immediately allow people who understand that they have a problem to skip the rigamarole and go straight to self-exclusion? Sure, Kalshi isn't a traditional sports book; it's not the counterparty to the trades and wagers that customers make, so they don't have the same incentives to hang on to losing customers exclusively. But it seems that they want to hang on to all of their customers. So why didn't NPR ask about that?

Of course, there's little reason to expect that a company will voluntarily take actions that will damage its own business model and earnings potential. The point of businesses is to make other people's money into their owners' money as efficiently as possible, after all. And Kalshi regarding the NPR story as a hit-piece and not wanting to be fully cooperative and transparent makes perfect sense, given that. But to simply say that Thomas' story was "cherry picked" misses the point; of course the people who wreck their savings by burning through tens of thousands of dollars on multiple occasions are the outliers. If that were the average outcome, there would have been immense public pressure for legislators to become more involved, and people would have been running on it as a campaign issue. And if the stories are rare enough to be "cherry-picked" then it shouldn't be much of a problem for Kalshi to allow those customers to leave.

Because note that the question at issue here isn't why Kalshi didn't detect that Thomas had a problem and bar him from trading sports futures on the site; it's why they didn't immediately allow him to block himself from the site when he came to them and said he had a problem. If Kalshi didn't see Thomas never trading again on their platform as a problem, why decline to allow for immediate self-exclusion and redirect people to "nudges?"

Now, maybe NPR asked those questions, and Dani Lever from Kalshi blew them off. It wouldn't be the first time a company simply ignored questions where the answers wouldn't make them look as good as they wanted to. But in that case, include that in the story, so that the audience has the complete picture?

Granted, it's not difficult to decide that, for all of Kalshi's high-minded rhetoric about their incentives, they're hard up enough for revenue that they keenly feel the loss of every customer who decides they need to cut themselves off, and this explains their reluctance to take problem gamblers at their word. But that's an inference that could have been better supported, with a bit more digging.

Allied

The "American Infrastructure Alliance" has formed to push for data center growth in Georgia, Indiana, Iowa, Ohio, Pennsylvania, South Carolina and Texas. Supposedly, the goal is to work with state and local governments on "responsible development" and, in so doing, head of potential moratoriums on new sites. (As opposed to the America's Infrastructure Alliance, which concerns itself with transportation, and the AI Infrastructure Coalition, which is a different group.)

Personally, what I was looking for in the Axios story on the outfit was some indication that they would come out against data center projects looking to have the elected officials they read into their projects sign Non-Disclosure Agreements. After all, if the point is, as International Brotherhood of Electrical Workers international President Kenneth Cooper says, "development that benefits workers, communities and our nation," then the developers should be willing to lay all of their cards on the table, no?

Personally, I suspect that the group isn't going to be as effective as its members may want it to be. Because the issue isn't really benefits... it's costs. Data centers have a very high level of perceived costs for the communities in which they are situation. Water usage may be overstated, but in places where the electrical grid needs to be updated or expanded to manage the extra capacity, power costs go up, and data centers are are often large, ugly and noisy. And while the labor unions that have signed on to the group are touting the jobs that will be created are mostly going to be temporary... it may take a small army of electricians to build out a data center, but it won't need all of them to maintain it. And if the union workers who build data centers live elsewhere, the impact of their wages on the local economy will also be short-lived.

The various businesses and unions that are backing this new alliance are going to have to make the case that the benefits to the communities where data centers will be built will match or outweigh the costs to them. But part of what allows businesses to be successful is often the ability to externalize costs (like not having to pay for training data), and the expectation that communities will "take one for the team" has already been articulated by some parties.

If the idea that this is simply another scheme to transfer wealth from rural communities to corporate shareholders and union executives take hold, this project will be dead on arrival. Considering the substantial head start that idea already has, this "American Infrastructure Alliance" likely has its work cut out for it.

PS: I went looking for a website for the group, to see what they had to say in their own words, but didn't manage to turn one up. Maybe it's a bit early for that yet.

Action Required

Has there ever been an industry rollout, from a PR perspective, sort of worse than AI? Which has sort of been: “It’s coming, there's absolutely nothing you can do about it, it might end the world, it could very well destroy your job. But you know, as a consolation, this chatbot will help you describe how to commit suicide.”
David French, The Opinions Podcast: “Trump's Political Instincts ‘Are Clearly Fading’.” 19 September, 2026

Some of this is that the companies have done a terrible job at messaging. They have effectively said, “We’re going to destroy all your jobs and potentially destroy humanity, but we’re going to IPO for a massive valuation and we’re going to set up camp in your local communities with data centers, which by the way, don’t employ people and drive up the cost of power. But, it’s okay. China.”
Jonathan Kanter, Decoder Podcast: “Does AI need an antitrust exemption so it doesn’t kill everyone????” 19 September, 2026

While Dario Amodei may think that the messaging around generative automation is less important than prior failures of corporate America to live up to its promises to make people's lives better, if this is the impression that people have of the public messaging from Anthropic, OpenAI, XAI et cetera, the companies sure aren't doing themselves any favors.

Back when Mr. Amodei was backing up one of his engineers in a debate with an investor on X, he noted that earning the public's trust would require more than promises to cure cancer; the cure actually needed to be delivered. But when I read Messrs. French and Kanter characterizing the messaging from the industry, neither of them mention curing cancer. And perhaps that's the problem that industry players have; their lofty promises aren't actually making it into the public's consciousness. To be sure, I'm not convinced that many people take the idea that generative automation will be able to come up with a cure for cancer(s) in the short term, but the fact that the pledge isn't even part of the discussion is telling.

And maybe that's because cancer cures have been promised for decades now, yet have never materialized, but maybe it's because it's not anything that anyone expects immediately, and so it becomes a matter of giving companies a lot now, in the hope of some benefit later. And the breakthroughs that the frontier labs are actually announcing aren't immediately useful to the general public. In other words, while it's impressive that an OpenAI model disproved a conjecture concerning the planar unit distance problem, what does that disproof do for the many thousands of people who have lost their jobs because the C-suite of their company decided that agents could do the work more cheaply or simply to free up capital for data center construction? Perhaps the lofty promises are being ignored because there's no concrete indication that the generative automation labs are actual working to bring any of them to fruition. Rather they're the typical technology smokescreens that corporate executives seek to hind behind when they ask for public subsidies for their shareholders and their own salaries.

But as Mr. Amodei pointed out, this is not, fundamentally, an advertising problem; rather, it's an action problem. And while I think that yes, finding a way to do away with the scourge of cancer will show people that the industry is serious about doing something good for people, there are a myriad of much smaller problems that could use solutions in the near term. It occurs to me that the rapid pace of frontier model creation, to the point where the industry claims that they cannot be sure their products are safe, becomes a tacit admission that the current models aren't really useful for much of anything. Chatbots can, allegedly, completely wreck a person's mental health, but they can't reliably shave even one-half of one percent from people's cost of living. They produce little, if anything, that makes life better for the average person. Even people who are more productive as a result of using generative automation in their day-to-day work aren't necessarily seeing their wages rise from it.

While it's not true that the technology moguls are the only people benefiting from this new technology (the trades involved in data center construction are seeing dollar signs), for the average member of the public, this whole enterprise comes across as high-risk, low-reward, buttressed by justifications that seem flimsy on their face. This gives the impression that the technology companies believe that they don't actually need the public's support. Sooner or later, I suspect that the broader public will contest that with them.

Fire In The Hole

There's a decent amount of Internet schadenfreude being aimed at Shopify CEO Tobias Lütke after a podcast interview in which he noted that while generative automation allows employees to create more, faster, when they don't then check that work, it becomes "slop grenades" (perhaps more commonly known as "workslop") lobbed at other employees.

According to a Business Insider article about the interview, "Lütke said AI is most valuable when it makes people's thinking clearer and more concise."

Okay, fair enough, but in the all-hands memo that was reported, he said that "Reflexive AI usage is now a baseline expectation at Shopify."

That's significantly different from "consider your tasks, and where generative automation can make you clearer or more concise, use it." The memo seemed to presuppose that humans simply couldn't match the clarity and conciseness of an LLM. And if people are expected to reflexively turn to generative automation to do work, and are being graded on being seen to do so the "slop grenades" are a feature, not a bug.

In the all-hands memo, it's noted that: "What we have learned so far is that using AI well is a skill that needs to be carefully learned by... using it a lot." Well, this is what everyone "using it a lot" looks like.

The problem, he said, is that the tools make it easy to pass along work that has not been properly evaluated.

From the memo that I saw reported, the message seemed to be "use the tools all the time and let the feedback mechanisms sort it out." There's no call for employees to evaluate their outputs themselves before sending them. Now, I'm going to admit that I don't know if I've seen the memo as it was sent to Shopify employees... screenshots purporting to be the message in question were shared to LinkedIn, and that's what I read.

In any event, it seems to me that Mr. Lütke is running into something that I remember encountering in my professional career... people don't do what someone tells them... they do what they hear. And a hallmark of good communication is narrowing the gap between those two things as much as can be managed. While Mr. Lütke may have intended for Shopify employees to be careful to ensure that their use of generative automation didn't simply create more work for people, when employees read that "In a company growing 20-40% year over year, you must improve by at least that much just to re-qualify. [...] That sounds daunting, but given the nature if the tools, this doesn't even sound terribly ambitious to me anymore," their focus is going to be on meeting that benchmark, especially given that the memo also noted that "We will add AI usage questions to our performance and peer review questionnaire."

Of course Shopify didn't intend for people to use generative automation willy-nilly and then simply lob the raw outputs over the wall to their colleagues. But Mr. Lütke's memo didn't make an expectation that employees would check those outputs to see if they actually provided value clear to anyone. And that's where the miss occurred. Employees internalized the implicit assumption that all automation was good automation, and left it to feedback from others to inform them if they were on the right track. But the sheer volume of material that an LLM can put out in short timeframes can make evaluating all of it difficult, especially for someone who is receiving work items from a number of people.

Perhaps the real lesson from this is that CEOs are people, too, and they're just as susceptible to believing that their words are accurate carriers of their intent as anyone else. And of finding out the hard way that they're mistaken.

Islands

LinkedIn differs slightly from other social media sites due to its general focus on networking (although one might say that nowadays, it's primarily about social selling and general self-promotion). A side effect of this is that people, at least those who aren't hoping to defraud someone, tend to use their real names.

This leads to an interesting facet of the site; when people complain about businesses, they tend not to name the business they're referring to. Since their comments can be attributed to them by potential employers who peruse their profiles, there is a particular style of "LinkedIn speak" that people tend to use. This is most visible when people are discussing their own layoffs, but it appears when other topics are talked about as well. When people address the topic directly, the rationale tends to be the same... they don't want to risk being thought of as a bad hire of one sort or another. And so they don't name names.

And this makes sense. What I'm not sure makes sense is the broader response to this from the overall LinkedIn community. Or, for that matter, the public at large.

People fear burning bridges with potential future employers. And this frees up businesses from having to fear burning bridges with prospective (and current) customers and employees. Given that businesses don't have to care what people think of them (because they won't be named) they can act as they choose. They can suddenly throw people out of their jobs, they can treat employment candidates poorly and they can even mistreat their customers. Because even if someone takes to LinkedIn to decry this behavior, the company won't be named.

Especially ironic in this sense are the posts that purport to be warnings to businesses of the potential reputational damage of bad behavior when they detail the behavior, but don't name the business that perpetrated it. After all, how can a businesses reputation be damaged when no-one is willing to say that they did it?

The perception of powerlessness begets powerlessness. People feel that businesses can punish them with impunity, and this prompts them to keep their heads down, and not do anything that might draw negative attention to themselves; even when they believe that a business has behaved badly. This allows businesses to retaliate with impunity, because even if people were inclined to hold them accountable, they often don't know who the bad actor is. People see the overall lack of consequences, and the cycle continues.

One of the things that I tend to harp on is the idea that the United States is not a unified polity. Big picture, it's made up of a number of different ethnic and racial groups, most of whom feel that they have their own interests that are independent of (if not at odds with) those of the society at large. Drilling down, the overall individualism of American society creates circumstances where people look out for themselves first (even at the expense of others) because they are convinced that everyone else will do the same. And so it takes some pretty remarkable events to inspire people to collective action, and even that unity tends to be tenuous and short-lived. This prompts a feeling of isolation and powerlessness, which can often become a reality of isolation and powerlessness. 

Extollation

Per Bloomberg, AI innovator Fei-Fei Li says Americans should be setting an example for the rest of the world when it comes to extolling the benefits of AI.
AI 'godmother': US skeptics have wrong mindset about the tech

Benefits for whom?

By contrast, many countries in Asia demonstrate a more positive attitude toward AI, according to Pew. Li chalks that up to a strong focus on the technology throughout early education in those regions.
AI Pioneer Fei-Fei Li Warns of Risks From Growing Tech Backlash

But a lot of counties in Asia, like China and Japan, are much higher-trust societies than the United States. And a greater focus on technology during early childhood education is not going to fix that.

This goes back to the conversation between Gavin Baker and Dario Amodei over the weekend. One of the points that Mr. Amodei made was that many people in the general public "always suspect that we are cooking up some new way to screw them over," and part of the reason for this is that companies, governments and the technology industry don't keep their promises to make things better for people.

Take Uber, for example. They promised a better way of getting a ride. Now I can't go from the airport to a hotel when I travel without giving up personal information, and Uber has been accused of monitoring the battery levels of people's phones to judge whether or not they'd be up to paying surge pricing. An unmitigated good, this isn't.

Or consider the following, from a recent episode of the podcast Decoder. Host Nilay Patel makes this observation, almost as a throwaway comment:

I had a partner at a very, very large consulting firm tell me they had done hundreds of AI deployments and less than two dozen had yielded positive ROI. I said, “Can you just tell me that on the record?” And he said, “Absolutely not. We want hundreds more.”

If businesses won't be honest about the results their enterprise customers can expect, because they want to keep selling lucrative, but non-cost-effective, services, why would anyone think that businesses would be more forthcoming with the public?

Senator Josh Hawley can open a Congressional subcommittee hearing with: "We’re here today to talk about one of the effects of this AI boom: the partnership between the AI industry and some of the biggest corporations in America to effectively scam consumers out of every last dollar they have in order to buy products that they need and rely on," without sounding like crackpot because people honestly believe that immiserating them for the benefit of executives and wealthy investors is the plan. It's unrealistic to presume that people going to extol the benefits of a technology that they understand is primarily intended to be a weapon against them.

There are a lot of technology boosters and techno-optimists who will happily pontificate about how generative automation, general artificial intelligence or what-have-you will make work optional and usher in a post-scarcity, utopian, society. All companies have to do to get there is receive a subsidy in the form of massive amounts of data from the public without needing to pay for any of it. I suspect that it's rational for people to be suspicious that a industry that claims that paying for its inputs will doom it will actually be ready, willing and/or able to share the wealth later.

Negativity Bias

I do agree that the public has a negative view of AI (and that this is a big problem), but I don’t think it is primarily caused by me or any other AI leader warning about AI’s risks.  I think it is fundamentally a crisis of trust. I think that ordinary people don’t trust companies, governments, or the tech industry and always suspect that we are cooking up some new way to screw them over.  The causes of this go back decades and AI is just the latest iteration of it.
Dario Amodei. Saturday, 15 August, 2026

I understand where Mr. Amodei is coming from here, but I think that I would portray the public as a little less paranoid. I don't believe that companies are setting out to "screw me over." Rather, I understand that the point of a business is to transfer wealth from other people to the owners of the business. Period. A business can set out to do just enough to maintain itself as a going concern, it can swing for the fences or pick a point in-between, but at the end of the day every business I deal with regards some or all of the money in my checking account as its rightful entitlement and seeks to find a way to transfer that to its owners.

And businesses have been known to go overboard. Back in 2013, Jamies Graigie, then the CEO of Church and Dwight, was railing against laundry detergent pods, which gained traction in the market when Proctor and Gamble (which had been working on the technology since the 1960s) introduced Tide Pods. Mr. Graigie's problem with Tide Pods was simple; they eliminated waste by providing precise amounts of detergent. But that waste translates directly into higher sales, and thus, to profits. Which is what shareholders want. More recently (two weeks ago), a Senate Judiciary Subcommittee on Crime and Counterterrorism hearing accused corporations of using information gleaned about users to determine their willingness to pay higher prices, and thus charge them more. To be sure, there was a certain amount of grandstanding going on, with Senator Josh Hawley (R-Missouri) noting in a press release that companies were out to "scam working Americans out of their hard-earned money," but it's pretty much understood that this is how businesses operate.

So, as I would put it, the public has a negative view of generative automation because they understand that the goal of companies is to use technology to increase their profitability, whether or not the technology actually makes the public better off. And this is a rational understanding, because the big shareholders in companies expect this... it's why they invest in new technologies in the first place. Mr. Amodei's post on X was a response to an investor, Gavin Baker. And one of Mr. Baker's critiques in the overall discussion was that bringing up the dangers of generative automation, or actual artificial intelligence, if that ever arises, is frightening the public, and is thus a driver of opposition. And Mr. Amodei's response is basically that what people are reacting to is the perception, honed over decades (with the help of people like Senator Hawley), that the business of business is to benefit people like Messrs. Baker and Amodei (and maybe even Senator Hawley), at the direct expense of the public at large.

We’re here today to talk about one of the effects of this AI boom: the partnership between the AI industry and some of the biggest corporations in America to effectively scam consumers out of every last dollar they have in order to buy products that they need and rely on.
Senator Josh Hawley

And while I don't buy into the talk of scams, since most businesses are not actively engaged in any sort of fraud, the fact of the matter is that money spends just as well regardless of how it was obtained. A business that makes $100 from providing good value for money can't do anything more with it than one that sells someone garbage or simply finds a way to provide nothing at all of use in return. And to the degree that money for nothing is a more efficient business model than money for value, it doesn't require an Economics degree to understand which option investors will find more attractive; especially in the short term. And this actively creates the sort of perverse incentives that Mr. Amodei understands erode public trust.

I don’t think that a glitzy marketing campaign with a positive spin (which some have advocated that Anthropic do) is the way to win back that trust — at this point, saying that AI will cure cancer is more a cliche than it is inspiring, and most people think it is deceptive.  The thing that will work is actually curing cancer.  I think by far the most accurate criticism of AI companies including Anthropic is that we haven’t yet delivered on our big promises to benefit the world.  That is totally on us, and I think it’s the criticism you should be making, instead of all this stuff about messaging and marketing.
Dario Amodei

I would go a step farther. Generative automation has to do more than simply "cure cancer" (which is a much taller order than Mr. Amodei's X post gives it credit for). It has to make that cure accessible to people who need it, rather than it simply being another way for "some of the biggest corporations in America" to extract money from people by leaving them with the option of watching family members die. If Anthropic creates a cure for cancer, only for Johnson & Johnson or Eli Lily & Company to monopolize it and use it to extract money from the public, the promise to benefit the world will still be considered undelivered on.

And here, Mr. Amodei's optimism may be showing. In a low-trust society, where many people feel that they can only trust themselves to look out for them, it's going to take more than technological advancement to win back trust. It's going to take showing people that they, and their interests, are important. But Mr. Amodei is certainly correct in his understanding "that a glitzy marketing campaign with a positive spin" won't accomplish that.

Nailed It

Generative automation seems to have become the new hammer; for the person who has it as a tool, any task in front of them requires nails. I received an e-mail to notify me of a new Windows promotion that Microsoft is running as "the ultimate college bundle." Buy an eligible Windows 11-base personal computer "for college" and receive: a year of Microsoft 365 Premium, a year of XBOX Game Pass Ultimate and a customizable XBOX Wireless Controller. Okay, fair enough. Not my thing, but I can see how it could be an attractive offer. What I don't see is how it relates to this:

The Slop is strong with this one.

Now, I'll be the first to admit that for me, college was quite some time ago, and I don't claim to be completely up on what "Generation Z" is all about these days, but I'm pretty sure that whomever prompted (and whomever greenlit) this picture isn't either.

Because while I suspect that there is an overlap between young female gamers and anime fans, I wouldn't think that the two were synonymous. And so this picture doesn't say "gamer" to me; it says "anime cosplayer." Because who else would go to their college library wearing a sērāfuku and bring a ninjatō with them (especially one that seems to simply float in the air behind them, as though looking over their shoulder)?

This usage of generative automation strikes me as absolutely perfunctory. While I understand Microsoft wanting some sort of image for the advertisement, why use a poor-quality generated one? Was a photograph of an actual college-age student, in a library or dormitory, with an actual book really that expensive? The time, energy and money to source something that resembles the target market was really that dear?

"AI Slop" has been defined as usage of generative automation that demonstrates contempt for the audience; the idea that since the audience will lap up literally anything placed in front of them, one may as well serve up the cheapest garbage one can find. That definition came to mind when I saw this. Nothing about this illustration says: "We understand young people who want to use their PCs for both school and gaming." Rather it conflates gamers and cosplayers and presents the sort of lazy stereotype that shows not only a lack of understanding, but a lack of desire to understand.

Missing a Step

According to Fortune magazine, Amazon and Microsoft are spending $400 billion on AI—and investors are low on patience. Patience for what, I wonder. The article notes that investors may be looking to understand which, between Microsoft and Amazon "soak up all that investment and spin it into gold faster," but it never gets into how this spinning would actually work.

And that seems to be the importance piece. If one presumes that there actually is an "AI Layoff Trap" as Professors Gerry Tsoukalas and Brett Falk lay out in their recent paper, there won't be much gold to go around, as the rush to replace people with automation grinds its way through corporate customers, crashing demand for most non-essential items. While people like Elon Musk have apparently convinced themselves that ubiquitous automation will rapidly usher in a post-scarcity economy, the techo-optimists, or at least the articles that lay out their positions, end to also be short on details as to exactly how this will come about.

I'm really bad at predicting the future, and so I don't claim to have much insight into how things will turn out. I've just learned to be skeptical of claims that the world will suddenly undergo radical change when the mechanism for such change can't really be identified.

For the hundreds of billions of dollars being put into generative automation to pay off, it has to create something that allows people to sustain themselves and an economy that relies heavily on discretionary purchasing. Right now, it doesn't appear to be doing that, and while hope that this will change may be common, that hope is going to have to be realized before there can be any sustainable return on that investment. 

Lego Lockdown

The Lego section of the Toy department of the local Fred Meyer store.

Locking merchandise cases are popping up more and more often in local stores. Whether they're for liquor, Lego sets, laundry detergent or underwear, the number of local stores that use them to prevent theft is growing.

When, that is, the stores even stay open. The Fred Meyer in Redmond, which I'd visited from time to time, has been closed, ostensibly due to a high level of shoplifting. It's the sort of story I've heard several times in the area... employees are told to not challenge thieves, for fear that they are armed, and so people learned that they could simply walk out with armloads of merchandise, and the store eventually simply closes.

This sort of thing is usually treated as a matter of corporate economics when the news media takes it on, but perhaps it should be covered as a policy story. The northern set of Seattle Eastside suburbs aren't exactly low income; there are enough Microsoft (and other tech company) millionaires in the area for McMansions to be fairly common. While poor people from other parts if the region have cars, too, and can drive to other locations to steal things, I suspect that there are a good number local people with an eye towards what they may be able to get for free. While the President chases supposed vandals of the Lincoln Memorial Reflecting Pool, the economy, and consumer confidence, continue to limp along.

And given the number of technology jobs that have been eliminated, so that Microsoft, Amazon and the like can pay for the massive capitol expenditures needed to fund the construction and operation of massive datacenters to power generative automation, it's likely that many people whose livelihoods were downstream of those tech workers are also suffering.

The Administration may be content to ignore what's going on in Blue states, in order to focus on the President's pet projects, but, sooner or later, something's going to need to be done.
 

Destined

Because it's the "in" thing, video game studio Bungie (the creators of the Halo franchise, and, more recently Destiny and the rebooted Marathon) has laid off a significant number of staff. According to a statement published online, part of the reason for the reorganization of the company was that "Destiny 2 fell short of expectations these past several years."

I played Destiny and Destiny 2 for quite a long time, even if I wasn't all that great at it. They were fun games, at least for a time. One day, after not having played Destiny 2 for some time, I loaded it up again, and within 30 minutes was having a grand old time, running, jumping around the map and shooting aliens in the face. After a couple of hours, when I logged off, I asked myself, "Why did I stop playing this game?"

The next day there was an update, and a new environment opened up. Being the sort of gamer who loves to explore the environments (especially when they are as well-crafted as Destiny's were), I jumped in, and found myself in a mission to fight my way through some sort of spaceship or orbital station... I don't remember which. It was fine, until my character encountered the final boss. And died. I tried again. And died. Over, and over, and over again, I adjusted my tactics, tinkered with my character's loadout, and tried again. Only to be killed by a powerful, and bullet-spongy final boss. Which reminded me of why I'd stopped playing Destiny 2.

Destiny 2 was a very particular type of game, and one aimed at a very particular type of first-person shooter player. It likely goes without saying that I was not that type of player. And so I eventually found myself pushed out of the game, because I didn't have the inclination, or the time, to mold myself into the sort of player that the game was geared towards, and Bungie wasn't ready and or able to make the game more accommodating of other types of players.

According to the "Bartle taxonomy of player types," first laid out 30 years ago by Richard Bartle, I'm an Explorer. I'm the sort of player who love to have their character wander around in new environments and just check them out. Finding a secret or hidden pathway to something I've never seen before is the highlight of a session for me.

And Destiny 2 wasn't built for that sort of thing. It had a number of really interesting places to explore, but many of them were gated behind difficult fight sequences, or were parts of raids and weren't really designed to just roam around and examine in detail. And so I drifted away.

With new companies, the question they have to be able to answer is: Why should people stop playing their current favorite games to play yours? For established companies like Bungie, the question becomes: Why should people continue to play your game, rather than explore what else is out there?" And I think that they didn't take that question as seriously as perhaps they might. And the business is starting to wither as a result.

Unneighborly

According to the Wall Street Journal, State Farm sales agents are up in arms over a scheme to reduce their compensation and benefits, now that State Farm has lost the title of premier auto insurer to Progressive.

I don't blame them.

I used to be a State Farm customer. I was with them for many years.

One day, my car was in the parking lot of a local drugstore, when some yahoo hit it and drove away. I called State Farm and was told to use their online claims tool. Which I did. A week went by, with no response. In the meantime, I learned that I really needed to have my car fixed, because despite the fact that it ran perfectly well, the fact that it was now down a headlight and turn signal meant that it wasn't street legal.

So I called State Farm. And was told to use their online tool. I told customer service that I had. The agent looked, and, with clear surprise, told me that the company had simply dropped the ball; for reasons they couldn't explain, my report was still sitting there, waiting to be picked up.

Things went downhill from there. If someone had told me in advance how poorly things would have gone, I wouldn't have believed them. It was an unmitigated disaster. And completely unexpected, given that State Farm wasn't just some random online-only insurance startup.

And so I switched insurers. My State Farm agent, who I really liked, sent me a letter, asking me to come back, and promising me a discount on premiums. I felt badly for him; none of what happened was remotely his fault, yet he was the one expected to grovel. And I told him: what I was looking for was from someone for State Farm corporate to get on the phone and say that they hadn't met expectations. One mistake can be chalked up to human error. When literally nothing works as promised, there's a process problem that needs to be fixed.

I understand State Farm forcing cuts on agents, due to falling revenues. But when I stopped buying insurance through them, it wasn't due to premium rates; it was the impression that I couldn't be sure that I would get the services I was paying for. I shopped around when I switched companies, and the rates were pretty much all the same; the only real way to reduce what I was paying was to drop certain coverage. Given this, I suspect that I'm not the only person who didn't find State Farm to be good value for money. Making the agents take it on the chin won't fix that.

Can't Buy Me Love

Given the fact that part of the reason why SpaceX wants to build data centers in space is that they understand that many communities find living near ones on Earth sucks, the idea that opposition is the result of a foreign influence campaign by Chinese interests seems to be a bit misguided.

Look, I get it, there's the promise of money to be made from ubiquitous automation, and boosters of technology have long held that the general public is simply too ignorant for their concerns to be allowed to slow progress.

But there's a reason why there are individuals who aren't going for: "This will be great for everyone; just trust us," and that's because time and again, promised benefits haven't arrived (or, perhaps more accurately, haven't been shared), and rather than supporters of technology taking the responsibility to make those who bore the costs whole, fingers were pointed until the whole thing blew over.

If LLMs were going to be such moneymakers, why did Meta and Anthropic feel the need to pirate books to train their systems, rather than just paying for them? I think that there are more causes to be suspicious of generative automation firms than their supporters may be willing to cop to.

The fact of the matter is that "Big Business" are not considered trustworthy by many members of the general public, and "Big Business" hasn't done a whole lot over the years to change that. Generally speaking, it's rare that someone supports something where they bear the costs, but the benefits go to someone else.

All of that said, I think that it's more than "the rich" and "tech moguls" who think (or want to think) they smell a rat. There's a reason why the term "Luddite," is more of a pejorative today than it was originally. When groups of people are pitted against one another, it's common for one or both sides to not only believe they deserve to win, but to see the opposition to them as illegitimate. And a lot of people have hitched their wagons to the generative automation star. It's not only the well-off among them who might resent pushback.

Consolation

 Microsoft's Brad Smith made a blog post today about responding to concerns, especially from young people, about generative automation and the possible impacts on employment.

It's a very cromulent bit of corporation-speak, designed to convey to readers that Microsoft isn't so wealthy, powerful or distant that it doesn't care about people's worries concerning technology. And Mr. Smith is willing to say some things that other corporate executives have been less willing to; he notes that there has been "corporate pressure to reduce headcount to help pay for AI’s enormous capital expenditures," rather than pretending that job cuts were the result of generative automation being able to do the work better. I think I still fault him for making this admission only after that reality had become public knowledge, but honesty is honesty.

But as I read it, I kept being drawn back to the Copilot Super Bowl ad (since removed from YouTube) that aired in 2024. For all that Mr. Smith says that the goal of generative automation is to help people do better work, rather than replace them, the promise of that initial advertisement was that people could be "soloprenuers" by letting Copilot do the things they would hire other people to do. And I said at the time: "If one's fear is that technology will drive isolation, or that AI will render one's skills obsolete, the message of this commercial is not reassuring." I would liked to have seen Mr. Smith acknowledge that there are people who are afraid because Microsoft's messaging gave them reason to be afraid.

Mr. Smith notes that in their book Open to Work: How to Get Ahead in the Age of AI, Ryan Roslansky and Aneesh Raman note five "soft skills that are uniquely human," namely: curiosity, creativity, compassion, communications, and courage. Which is all fine and good. But these are not traits that are in short supply. If the majority of people thought that the demand for these skills were high enough to keep the job market afloat, there wouldn't be so much worry about a jobpocalypse on the horizon.

If widespread adoption of generative automation by businesses is inevitable, and the jobs shed to free up budget for capital expenditures aren't coming back, the technology is going to have to create demand for human labor somewhere else. And "human observation and insight" aren't going to be that demand unless, somehow, the percentage of returns going to labor increases markedly, such that demand for automation-produced goods and services really skyrockets. And that's not a prediction that technology industry analysts seem to be making at this point.

To the degree that progress can be described as a process of creative destruction, I suspect that the public at large is better off when creativity drives destruction, rather than destruction demanding sudden creativity. But the public at large isn't calling the shots. It's the investor class, in search of returns on their investments. And as a high-ranking executive at a publicly-traded company, it's not difficult to tell which of those groups Mr. Smith is more immediately answerable to. And so I wonder about the motivation of this post, from a public relations standpoint. Is this honestly what the leadership of the company is thinking, or is it being conciliatory in an attempt to head off anger that may impact profitability and investor confidence?

Early

I was at Costco this weekend, and noticed that a giant skeleton decoration was already up and on sale. While June seems pretty early for Halloween decorations to be on store shelves, it's something of a necessity if things are both going to be sold, but not in the way of the earlier and earlier start to the Christmas shopping season.

Personally, not being a big Christmas person, I'm not a fan of the longer and longer time that's being devoted to it. But I get it from the point of view of retailers. To the degree that Christmas shopping can be chalked up to impulse buys, it makes sense that stores that have items out first are going to capture sales. And this sets up something of a race to have things out early. And that pushes other seasonal items, like Halloween, earlier in the year.

We'll see if Halloween decorations are on the shelves in May next year, or if the march of the start of the Christmas season takes a year or two to catch its breath.
 

Promptly

With the understanding that I can't validate that this is even legitimate, this is another of those things that popped up on LinkedIn for people to have a good laugh at. It strikes me, however, mainly as weird. Sure, on the surface it's yet another "someone meant to have generative automation write something, and wound up sending the prompt, instead," but the prompt itself seems off to me.

"A warm but generic rejection email that sounds polite yet firm."

Don't companies have those? Who's actually expecting something other than a form letter? Why craft a new "generic" message for each rejected candidate? Isn't reusability the point of "generic?" This gives the vibe of using generative automation for its own sake: "We need to burn compute on a triviality to show that we're 'AI-forward'."

"Do not mention specific reasons for rejection."

I understand the rationale for this part of the prompt, but it still strikes me as risky. After all, there are likely non-specific reasons for rejecting a candidate that generative automation could come up that would still be a problem, if they aren't related to the job at hand. This is something that it strikes me that one would want laid out beforehand, for just that reason.

"Make the candidate feel like they were strongly considered even if they weren't."

Considering that the automation likely wouldn't know one way or the other to what degree a candidate was considered, I can understand having it default to implying that everyone was strongly considered. But I'm not sure that it's a good idea to have LLMs tell people something that may not be true... Once it's considered legitimate to have generative automation mislead candidates, even to spare their feelings, I'm not sure how one keeps people from asking the LLMs to deceive other stakeholders. And I'm not sure it takes much imagination to see how that starts ending badly, especially if the automation starts telling outright untruths.

"Remember to use the candidate name and company name variables."

Why is the company name a variable? Does it change somewhere along the way? This gives me the impression that this is coming from a third-party recruiter, who works with a number of different clients. I suppose that a holding company could have a lot of smaller companies under its umbrella, and centralized HR for all of them, but given that the company name shows up in other parts of the e-mail, it doesn't seem necessary to call it out again. And again, why not use a form letter? There's nothing in the prompt that calls for any candidate-facing personalization from their résumé or cover letter. I'm not sure what just using their name is supposed to do.

Of course, the fact that a prompt was sent to a candidate who was supposed to receive a rejection message means that messages aren't being vetted prior to being sent. Which makes some sense... after all, generative automation is supposed to be able to handle all of this. But even the prompt screw-up aside, if the idea is to generate responses to candidates on the fly, it seems that it would be wise to have something that checks things before they go out, if only to make sure that something entirely random didn't find its way into the message.

The final thing that stood out to me was the redaction. I understand why the candidate wouldn't want their name out there, but blanking out the company speaks to a fear of retaliation that I'm not sure is healthy. It's not like there's something in this message that points to anything criminal, or even unethical... a prompt was screwed up along the way. If pointing that out publicly is the sort of thing that would lead an HR department to blacklist someone, maybe we as the public (and yes, I include myself in that) need to start having higher expectations of the businesses we give our money to.

Differentiated

It's been some time since I've used a graphic for The Short Form, mainly because, as I've mentioned before, it's hard for machines to read the text in a picture. But I was talking with an acquaintance earlier this week, and this part of our conversation stuck with me.

There is a difference between preventing bad outcomes, and preventing them from happening to oneself.

I suppose that it's an obvious sentiment, but I don't know that it's thought about all that much. In a lot of ways, it's like the difference between using The Club, and installing a Lojack, or other locator system, in a car. The Club is an obvious theft deterrent; it's goal is to not only make it more difficult to take the car, but to be obvious about that fact, so that the would-be thief moves on. But it doesn't really change their incentives; they simply look for a car that doesn't have such a device, and attempt to steal that one.

LoJack, and other locator systems, on the other hand, while being inobvious, carry a much greater risk to the thief if they do, in fact, steal the car... after all, it can be tracked by law enforcement, and that leads to a higher chance of being caught in a stolen car. But the fact that one cannot tell by looking if a car is equipped with a locator means that taking any car in a neighborhood where they're known to be in use carries higher risk. And this is why these systems often carry discounts in insurance premiums, they lower costs of insurers more broadly, and it's worth passing some of that savings on to those to have the systems installed.

This all came up in the context of the supposed generative automation apocalypse that's coming for certain sectors of the knowledge workforce. While a lot of people are offering various advice, from learning how to supervise automated systems to dumping the industry entirely and shifting to skilled trades, the general viewpoint is the same: This is going to happen, here's how you take care of you. It's modeled on The Club... a car is going to be stolen; this is how you ensure that it isn't yours. But maybe a LoJack model, trying to head off the worst of the transition in general, for everyone, would be better for all involved.

This Side, That Side

"Long term, you tend to remember that kind of negative branding," [University of Alabama Marketing Professor Karen Anne] Wallach said. "And negative language then becomes part of what you associate with the brand."

The tech startups NPR spoke with for this story said they understand the risks of alienating large numbers of people with their cryptic ads. But the upside is too great.
Do you understand this billboard? If not, that's the whole point

While this might seem to be just another story about tech, and how it divides people into groups, the above points to something important about in-group and out-group signalling. Sometimes, alienating the out-group is what the in-group demands. Groups, in general, are defined both by who is a member of the group, and who is not. And for groups that want to maintain some sort of claim to exclusivity, who is kept out can be much more important that who is let in. And hurt feelings on the part of those kept out be damned.

For technology startups who are not attempting to sell themselves to the general public, the idea that the general public is unwelcome can be just the sort of thing that their intended customers want; because it not only sorts, but stratifies. And sometimes, nothing sells a product or service like the idea that being a member of the target audience is proof of one's own superiority.

If an advertiser is willing to accede to an expectation of flattery, even at the expense of others, on the part of the in-the-know, clearly neither the advertiser, nor their audience, expects that any hard feelings on the part of the out-group will be a problem for them. And this is nothing new. I would submit that it's been a facet of human history for as long as there has been history. That said, it doesn't make the practice any less toxic, especially in its more strident forms. But perhaps that's the problem; toxicity has become such a common part of people's everyday lives that it goes unnoticed.

To Be Divine

Superhuman Platform, Incorporated, the company formerly known as Grammarly, is facing a class action lawsuit over a feature it rolled out at the end of the Summer called Expert Review. Expert Review, which was recently removed, was effectively a "this person would make these suggestions about what you're writing," sort of feature, and claimed to offer advice from virtual versions of people like Stephen King, David Abulafia and Julia Angwin (who filed the lawsuit).

When Superhuman Platform CEO Shishir Mehrotra posted an apology for the agentic feature on LinkedIn, he noted "valid critical feedback from experts who are concerned that the agent misrepresented their voices." When Ann Handley, who identified herself as one of those experts weighed in (before commenting on the post was closed), her primary complaint was "building a commercial feature around experts' names and reputations without asking permission, without notification, and without compensation." While Mr. Mehrotra claimed that "the agent was designed to help users discover influential perspectives and scholarship relevant to their work, while also providing meaningful ways for experts to build deeper relationships with their fans," given that it was a subscription feature, and Superhuman Platform wasn't sharing any of the money, it seemed more like they'd simply found another way to have people work "for exposure." And there's a reason why an increasingly common response to that sort of offer is "Fuck you; pay me."

As a random layperson, the whole thing strikes me as openly unethical; but entirely sensible. If generative automation is a race, and losing carries serious, or even existential, consequences, the time to be ethical is later. Ms. Handley calls Mr. Mehrotra out for an ethos of "take first, apologize later." And while I suspect she's correct in that, it's just like any other instance of "ask forgiveness, not permission;" permission wouldn't have been forthcoming, but forgiveness will be. And this is a rational presumption to make; Uber's known flouting of laws hasn't resulted in the general public deciding that the company is too untrustworthy to do business with. And it's unlikely that the Court of Public Opinion will render a different verdict for Superhuman Platform. Investors, on the other hand, are quick to flee a company that's unwilling to do what it takes to make itself more profitable, and they bear none of the risk for the actions the company takes in pursuit of those profits. It's not like anyone is going to spend time in prison over this, and even if someone were, it wouldn't be the investors; so why wouldn't they push for companies to place profitability over ethical considerations, given that it's unlikely that people and businesses with Grammarly subscriptions are going to go elsewhere.

The only way to stop companies (and people for that matter) from preferring to ask for forgiveness rather than permission is to be consistently unforgiving, regardless of outcomes. And that's a hard sell in a culture where many people's primary focus is their own sense of (or concern for) poverty. People may be angry when someone cheats them to pass the savings along to someone else, but they're often ready to look the other way when the savings are being passed along to them. And businesses know this, their executives are members of the public, just like everyone else. They may often speak in the stilted language of finance and investment, but they're not aliens.

Some heads may roll over this; if he's unlucky, Shishir Mehrotra's will be one of them. But Superhuman Platform, Incorporated will survive. People and businesses will still pay to use Grammarly, and investors will still see returns. And that all but guarantees that "take first, apologize later" will remain the standard order of operations.