Survey Says is a weekly series rounding up the most important polling trends or data points you need to know about, plus a vibe check on a trend that’s driving politics or culture.
The billboards started appearing last year, first in San Francisco, then elsewhere. A young woman’s face, normal but for her glowing eyes, stared out at us. The text beside her read, “Stop Hiring Humans. The Era of AI Employees Is Here.”
The company that put them up, Artisan, said the ads were merely meant to generate attention for its “AI-first” sales platform. In that way, the campaign was successful—here I am, writing about it—but in another way, it unintentionally reveals how so many of these artificial intelligence companies see human beings: as a vestigial organ, once useful but no longer, the only thing out of place in the otherwise perfect body of a corporation.
But in this supposedly shining future of a largely AI workforce, where do the people go?
This past summer, in a blog post about his billboards, Artisan’s cofounder painted a utopian vision. He imagined widespread AI adoption giving way to four-day workweeks, then to AI-supported universal basic income, allowing humans to never have to work again.
The naivety here is obvious. For one thing, he seems to think every job exists on a laptop. For another, in what world would the Republican Party, propped up by corporate cash, agree to tax corporations at high enough rates to send nonworking Americans regular payments that would cover every expense in every month, plus enough for them to buy cars and houses, take vacations, and so on? In what world would even the Democratic Party agree to that?
If only the naivete stopped there. Artisan’s cofounder, who is in his early 20s, also expressed surprise that his ads generated “1000s” of death threats, some of which he deemed “actually pretty funny.”
The thing is, the vast majority of people aren’t laughing.
Only 10% of Americans are more excited than concerned about the growing use of AI in everyday life, according to a recent Pew Research Center study. And the more Americans learn about AI, the less they’re excited. That same figure has been cut nearly in half since 2021, when it was 18%.
Meanwhile, 50% of Americans are more concerned than excited about AI—a number that’s grown 13 percentage points since 2021.
A primary reason behind that high anxiety is the very issue Artisan is glibly joking about.
Two-thirds of Americans think AI will eliminate more jobs than it will create, according to a Marist University poll conducted in July. And lest you think most folks are just afraid of new tech, data reveals that using AI more often does not mitigate those fears. Over half of millennials and Gen-Zers use AI tools often, and over half of each age group thinks AI will slash the number of jobs on net. In fact, 70% of Gen Z holds that view.
The public’s wariness of AI extends far beyond the job market. People think it will tamper with how we understand the world. Only 10% of Americans think AI will have a positive impact on the news people get over the next two decades, and only 7% are unconcerned about AI providing people false information.
Broadly speaking, Americans don’t see many upsides to AI. Very few think it will help people solve problems (29%), make hard decisions (19%), or think creatively (16%), according to another Pew study. Only 5% think AI will help people form meaningful relationships. Half of Americans think AI will make it harder to do so.
And at least two AI companies seem more than happy to usher in this new era of the incel. After years of promising AI would “fix” climate change and “cure cancer,” OpenAI CEO Sam Altman announced his ChatGPT service will, by year’s end, allow adult users to sext with chatbots. OpenAI joins Elon Musk’s xAI, which earlier this year launched AI girlfriends for his legion of pick-me losers.
But as much as people want to have sex with a model, they don’t want that model to be large language. A shocking 88% of Americans say they can’t imagine ever having an erotic or sexualized relationship with a chatbot, according to a recent YouGov poll.
There is an undeniable air of unseriousness that hangs around AI companies. They make grand claims about fixing (or ending) the world, then go all-in on porn.
Despite OpenAI’s potential $1 trillion valuation, 66% of American adults have never used its flagship product, ChatGPT.
Despite this, the sector is rapidly becoming a pillar of our financial system. One estimate says AI companies are responsible for 80% of stock gains this year, no matter that virtually none of them are making profits proportionate to their worths.
Look at OpenAI. The company is preparing for an initial public offering at up to a $1 trillion valuation. That would put it at nearly five times the value of the McDonald’s Corporation, in spite of the fact that OpenAI currently makes less than half of the fast food giant’s annual revenue.
Really think about that: McDonald’s has over 44,000 locations across more than 100 countries. It has among the most recognizable logos in the world, and it sells millions of items every day. Meanwhile, 66% of American adults have never used ChatGPT.
An AI hype-man might say that comparison is inapt since a company need not have many ordinary users to justify its large market cap. In that case, the numbers are about the same as compared with RTX Corp., the largest defense contractor in the world by market cap.
Unfortunately, it gets only more absurd. AI chip-maker Nvidia recently became the world’s first $5 trillion company, and while its market cap has slipped some since then (due to fears over AI companies having bloated valuations), it is still worth more than 21 times as much as McDonald’s. And yet its revenue is just five times that of the burger chain—or about 1.6 times that of RTX.
Or, if you want something truly ludicrous, Nvidia is worth over five times as much as Walmart but pulls in less than one-fifth of Walmart’s revenue.
But it’s not just that these AI-related companies are likely far overvalued. It’s that very few are profitable and are instead kept afloat by steady infusions of venture capital banking on the idea that one day, hopefully soon, the investments will pay off. And then there’s the problem that 95% of organizations using AI programs are getting “zero return” on their investment, and that some AI companies say that this high failure rate is actually good. And that the companies are incestuously investing in each other. And that all of this is clearly a bubble, as major news outlets are finally starting to acknowledge.
All of this danger is hiding in plain sight, too.
Visit the website for Artisan—the company running those pro-layoffs billboards, if you already forgot its generic name—and you’ll find a banner at the top. It isn’t about their product; the page it links to does not once include the words “revenue” or “earnings” or “profit.” Instead, it’s about how the company raised $25 million in Series A funding.
It’s a banner to hype the company to future investors. It isn’t for you.
Any updates?
-
This past Tuesday, Zohran Mamdani became the mayor-elect of New York City, a development that was basically unimaginable earlier this year. Before he won June’s Democratic primary, the polls had him trailing disgraced former Gov. Andrew Cuomo—but the polls were wrong. So how’d they do in the general election? Quite a bit better. A Daily Kos average of polls since the start of August showed Mamdani with 46% support, and as of Friday, he has 50% of the vote in the election, meaning the polls were just 4 points off. The miss was bigger for Cuomo, though: Polls showed him with an average of 29% support, but he scored about 42% in the election—a difference of 13 points. Cuomo beating his polls likely has a lot to do with Trump’s last-minute endorsement of his former foe, which seems to have hurt Republican Curtis Sliwa’s performance. Polls had him with 16% support on average, but that fell to 7% in the election.
-
Trump wants to start blowing up nukes again, testing the weapons for—well, it’s not clear. Either way, Americans are broadly skeptical of it. Only 33% support resuming nuclear weapons testing, according to the latest YouGov/Economist poll. Forty-eight percent oppose resuming testing, while another 18% are not sure. For what it’s worth, a strong majority of Americans (64%) think the invention of nukes was a bad thing, per another YouGov poll.
Vibe check
Gerrymandering is bad, but it’s worse for Democrats to disarm while Republicans keep at it. And voters in at least one blue state agree.
This past Tuesday, California voted to allow the Democrat-led state to redraw its U.S House districts. The move pushes back against Republicans in other states who are further gerrymandering their maps at Trump’s behest.
The trouble is, it’ll take much more than just California fighting back.
Due to recent redistricting, Republicans are currently on track to pick up four House seats on net, according to a Daily Kos analysis of new House maps that are in place or have passed a major legal or electoral hurdle, such as in California.
California’s impending redraw is expected to flip five Republican-held House seats, but that merely offsets GOP gains in Texas, which redrew its map in August. Since then, Republicans in Missouri, North Carolina, and Ohio have redrawn their maps to further advantage their party. And even more GOP-led states, such as Florida and Indiana, are moving to do the same.
Utah is more complicated. It’s also redrawn its map due to a court order, and while its new map gives Democrats a better shot in two districts, Republicans are still favored in each. That said, if next year’s midterm elections are a blue wave, the map could backfire on them, potentially giving Democrats two seats in the state, where they currently have none.
If these maps pan out as expected, and no more are passed, Democrats would need to flip seven seats on net to retake the House.
Virginia, your time is now. Maryland too. And Illinois. And New York.
