LBL #10: AI billionaire accelerationists perform alarm about a permanent underclass

AI billionaire accelerationists are wearing such heroic furrowed brows over the poverty their wealth-creation will cause! Yet I cannot find it within myself to swoon.

Last I checked we already have a permanent underclass created by tech disruption. What else would you call the millions of contract workers, ride-share and delivery drivers who live at the mercy of volatile platform fees without labor protections and benefits?

How can Silicon Valley be “bracing” for the underclass, as asserted by this insider-y-ish NY Times guest op-ed, when they are the architects of it? The underclass, if it is anything, is the semi-permanent labor force that lives and works within the intricate cage of apps and algorithmic surveillance that Silicon Valley not only designed but spent billions to defend in courtrooms and statehouses everywhere.

And now after decades of blowing up labor laws to satisfy venture return expectations, we’re to believe that they want to eliminate work and create lives of endless leisure by redistributing AI wealth? I read and reread both the NY Times piece and Anthropic CEO Dario Amodei’s essay and couldn’t find anything signaling a desire to genuinely engage with the political mechanics of antipoverty measures and unemployment.

AI accelerationists are entering their performative male eras, and there’s no need to stand and applaud. It’ll pass. Happens to every billionaire eventually. There’s new panic because the destructive economics of tech wealth are starting to harm engineers and coders. But I’ve spent much of my career as a journalist in Los Angeles watching those same economics hollow out the block, squeezing brick-and-mortar businesses to the brink.

Algorithms have allowed factories in Shenzhen and drop-shippers in the U.K. to compete for the business of selling gifts, Joss paper and red envelopes once exclusive to Chinatown. My first stories as a journalist were for patch.com about the extinction of the taxi-cab industry in Los Angeles, which, for a decade or two, was a lifeline for Armenian and Iranian immigrants fleeing upheaval. Those cabbies drive for Lyft and Uber now, but if Waymo catches on their livelihoods will be stolen twice by tech disruption. There was no fund then, and there will be no fund now.

It’s disappointing that Amodei’s essay skips this recent history to instead theorize about how AI’s will affect widget production and the way farming equipment changed agrarian societies. The piece is 22,000-plus words of wobbling speculation larded with parentheticals, a sci-fi novel with the soul of an investor prospectus and all the style of a dictionary. Most of it is future-casting about AI’s massive, destabilizing threat to humanity - which of course Anthropic is uniquely equipped to defend us against. Just a few paragraphs directly address AI-related mass unemployment. And one of them contains this sentence:

“We are thus at risk of a situation where…AI is affecting people with certain intrinsic cognitive properties, namely lower intellectual ability (which is harder to change). It is not clear where these people will go or what they will do, and I am concerned that they could form an unemployed or very-low-wage “underclass.”

Amodei writes about people who are dumber than him as if they are underperforming assets to be reassigned. But in the future, people who are dumber than Dario Amodei will do as we have done in every age: we will eat, breathe, struggle, seek love and meaning and generally go about our lives enjoying rights that do not change based on the market value of our labor.

You don’t have to believe him to conclude that the shows of conscience in his January essay have been highly lucrative. A month after its publication, Anthropic’s $30 billion funding round left the company’s valuation at $380 billion, up from $60 billion in March last year.

Strange that being performative is so profitable. Amodei and other Anthropic founders have promised to give away 80 percent of their wealth. But if you plan to capture some large proportion of the world’s entire economy, isn’t keeping 20 percent of it for yourself unfathomably greedy?

Here’s a prediction: before generous billionaires decide to gift us all with lives of endless ease in which no one works while AI agents do everything in the background, we’ll have an economy in which only a few people can pay for that privilege. We’ll face a political environment in which a powerful minority will try to block any policy that places more money in the hands of the poor. And in this future no one will be handed anything, because that minority hates the very idea of redistribution and has built their political identity around it. If that sounds familiar, it’s because that’s the future we have now.

Even the most free-thinking of the techno-optimists are ending their flirtations with redistributive socialism. Just 2 hours drive from Anthropic headquarters in San Francisco, Stockton’s Universal Basic Income experiment ended in 2021 despite promising results. Former Twitter CEO Jack Dorsey promised $1 billion to the cause but his foundation has not since issued another UBI grant. In April, the one-time apex predator of Silicon Valley’s performative males, OpenAI CEO Sam Altman, told the Atlantic he “no longer believes in UBI” as much he once did. (If you’re curious OpenAI’s performative tech manifesto/investor prospectus is even worse, a 13-pager beginning with “Let’s Talk” in 48-point font)

Amodei discusses poverty and inequality more broadly in his previous essay, “Machines of Loving Grace,” a breezy read at 14,700-plus words. But even here Amodei prefers to discuss inequality as a grand global struggle, except for one bullet point, labeled “inequality within countries.” The only answer he offers is to the pesky problem of “inequality within countries” is to trust markets to bring down the cost of high-value technologies.

I do trust markets: to create paywalled products for which the monthly subscription fee steadily rises over time in a nontransparent way. Last month Anthropic removed access to Claude Code from its lowest paid tier amid user reports that the token cost of various computations has risen.

What would it look like if Silicon Valley care about poverty? If Claude becomes “a country of geniuses in a datacenter” as Amodei keeps threatening, at what point will those geniuses decide to take on the issue that in 2020 just 22 cents out of every dollar given to Temporary Assistance for Needy Families actually made it to needy families? How would exactly would an army of Claude agents ensure that progress in easing the poor’s access to public aid isn’t erased each time a conservative congressional majority gains power? If these AI founders are so public-minded and their models so beneficial tor society, why can they do nothing for the blocks and neighborhoods their headquarters are located in? And why are the employees who built the foundation for all of these billions so scared for their livelihoods?

AI accelerationists want us to believe that they are creating a future in which they will make our problems obsolete. But we already know the future they’re building because we live in it. That world contains both work and poverty.

If the accelerationists want to eliminate poverty to prove the worth of the billions of venture capital they’ve monopolized, I’m all for it. But I don’t think I’d want to live in a world without work. Not because I love capitalism or getting sweaty, but because in every era and age, having something to work for is what makes life worth living.

Work isn’t just a set of errands performed for benefits and salary. It’s also meaning, connection and identity. We work to make great art and realize dreams and ambitions; to afford better clothes, homeownership or a family; to feel a part of something bigger, to maintain important relationships and form new ones.

None of these are problems human society needs to solve. They are exactly the parts of my life that I would never give up or automate, the parts that let me know that I’m alive.