I don't think there's any quick fix. You have described how the tech Bros have dug their own PR graves with great flair and efficiency.
I think AI will be rehabilitated in the eyes of the American public when really wonderful things like cures for cancer come out of it. It will have to prove its usefulness.
I also think that the AI development path suffers from gigantism— the conviction that simply building the database of everything is the way forward. I think smaller, dedicated AI tools that consume less compute, energy, and other resources are the way forward. It is true that you can peel a potato with a 10-foot knife, but there are more efficient ways to do it.
Ben, the real shame is that Google and other major data center builders learned in the early 2000s how to reduce power use and operate more efficiently. Many of today’s AI data centers are going up so quickly that they aren’t always taking full advantage of solar, thermal, and other proven approaches developed over the past 20 years.
And when local or state governments eliminate or sharply reduce taxation, you can see where some of the public pushback comes from.
While there are many small-scale versions of the rot that we could point to (everything-as-subscriptions, algorithmic marketplaces squeezing both sides of the market, ad models incentivizing ragebait and addiction), I think most of it boils down to the requirement to grow at any cost.
This is not, however, universal. In the old days when Google and Microsoft were racing to see who could build the largest search index for their search engines (a more innocent time), eventually Google realized that there were seriously diminishing returns to a larger index compared to the cost to support it. Today Google's search index is basically hard-capped at a specific size, and "least recently used" sites are evicted, which explains why it's impossible to find websites from the early 2000's through search.
In my time in the datacenter world, I asked some Google VPs around 2018 or so, "At what point will Google stop growing its datacenter fleet?" They were unable to comprehend the idea. It has only gotten worse.
Partly I think that this attitude will only be changed by active antitrust enforcement. Even this morning you hear executives saying things like "the anti-datacenter thing is just a phase," because their stock price depends on them believing it. Breaking up a company is the only way to get through to the MBAs that they need to change.
And on the more grass-roots side, we can't get people off their tech dependence without actively helping them. And that requires us to get out there and talk to people about their tech problems, and then help them fix them.
For example, my daughter's daycare had their website enshittified by their hosting provider, and they were required to pay $100/mo to keep it online (and they had no idea how to update it). I redid their website as a free static site. I talked to a teacher friend and helped them write some auto-hotkey scripts to make grading student work easier. I helped another friend use AI to automate pulling stats from websites that hosted their content into a portal they control. I helped another friend deduplicate and compress decades of data they had managed ad hoc over the years. Another compress the images on their website to reduce bandwidth costs.
In doing this, people see tech as a useful tool that they can harness, and it becomes much easier to separate the technology from the extractive company that leeches on their wallet and attention. The fact that most of my friends in tech aren't doing stuff like this (outside of family members) and the fact that most people have to be prodded a bit to open up about their individual tech problems, I wish I had a simple fix for.
Robots can't create surplus value, but they can sometimes create consumer surplus. Automation makes it easier for people to survive but also makes their lives feel more worthless. Short of "destroy all the capital stock in a WWII style orgy of destruction" I am not sure there actually is a solution.
The tech CEOs and thought leaders have so much contempt for government, voters, and the public at large that they literally never thought they could do anything to limit them in any way. The only PR that matters is PR to investors and other business executive customers. So the "AI will put everyone out of work and usher in a permanent underclass" line they've been slinging for years now made sense as a sales tactic.
Pretty telling that the datacenter backlash in the last year is making Altman and Amodei suddenly backtrack and say "actually maybe all white collar jobs won't be replaced..."
> I’d love to hear your thoughts on how the tech sector can clean up its image and make a positive, populist case for itself.
There needs to be a radical shift in how out industry manages diversity. It's maddening that the people who will be affected most by the current gen-AI scourge have effectively zero representation in industry or academia. The real answer is that we need to burn capitalism and the ideas that bolster it to the ground, but that seems to be even more of a pipedream than my current proposal.
*Are* people being made miserable by AI? How much of the hostility comes from experiencing the technology, and how much from what it has come to represent: tone-deaf founders, their politics, and broader economic insecurity while the industry vacuums up billions in investment?
I mean, really, how often do most people encounter AI in everyday life, and how often do they even recognize it as AI? I reckon good uses of AI often go unnoticed, like good CGI is often invisible. But ok, other than through chatting with me, my parents for instance only ever see AI through the Google summaries at the top of their search, some generated images, and occasionally in a product/tech support chat, and they are probably more representative of the average person than chronically online people like me. I have a hard time believing those encounters alone could explain such hatred.
I don't have data on this, but my guess is that much of the hostility is rooted in resentment toward the people and circumstances surrounding AI's development, amplified and reinforced by news and social media.
> How much of the hostility comes from experiencing the technology, and how much from what it has come to represent: tone-deaf founders, their politics, and broader economic insecurity while the industry vacuums up billions in investment?
I get what you're saying, but isn't it the case that the purpose of a system is what it does? Could we have gen AI without the foundation we describe (as in, does that make historical sense)?
How about: take seriously (or at least read) the social critiques of technological society that date back to the Industrial Revolution and dawn of quantification/optimization?
I have a mass psychology-based explanation for data center hate. It’s largely unfalsifiable and based on personal anecdotes, but hopefully people find it interesting.
Technological innovation has, since the Industrial Revolution, run counter to the immediate interests of the masses in a market economy. Part of the American ethos, however, was that we Americans take a gung-ho attitude towards the new industries that are opened up by new technology, hoping to jump on the bandwagon and see success. Over the long run, this doesn’t work for the majority of people. This is the observation that created the first progressive era.
However, 20th century attitudes towards communism created an environment in which this observation becomes stifled in the mind of the average American. Complaining about the decline in real wages is fundamentally un-American to the masses, because we believe in the free market. In red states, among the most communism-hating sectors of the population, they watch from a metaphorical cuck chair as Musk develops new rocket technology. They cheer him on despite neither being involved nor directly benefitting. Moreover, they cheer him on despite the fact that the market economy which allocated the capital for these developments has left them behind.
Now let’s consider AI. There is no hope, even theoretical, that someone in the boonies of Texas will become involved in the industry behind AI. It is so powerful and all-encompassing that anyone outside of the industry worries about being replaced and becoming superfluous to those in the industry. Far from the days where Ford required a large number of factory workers to build an empire, for the first time we see a technology that can build empires almost without anyone at all, except for a tiny minority of highly educated elites mostly residing in Silicon Valley and traveling unionized construction workers. Those left out rightfully feel hopeless. They should’ve started feeling that way long ago, but now it is undeniable.
However, a religious belief in the free market is still in their DNA. They will never complain about being replaced publicly, or even to themselves. They will instead funnel this anger into a much more American form of complaint: NIMBYism. No data centers in my neighborhood!
As for what tech people can do, I think we have a much larger issue than a PR issue. We have a structural issue. Capitalism is reaching its natural conclusion as new tools emerge to automate everything. We as a society need to agree on a system that distributes the benefits of an autonomy-dominated world to a population much larger than its market economy can support. And we are utterly fucked if we continue on our current path. These elections are a small step in the right direction, and historically large steps in the direction of socialized economies don’t work well—I’m hoping for many more small steps as we approach the final hour.
As to the fix, in my view the whole thing needs to be re-thought and re-implemented with reliability and trust-ability at the core. Starting with the training data + labeling, maybe it could handled in a more open, collaborative, manner; perhaps more like Wikipedia than GTA.
I don't think there's any quick fix. You have described how the tech Bros have dug their own PR graves with great flair and efficiency.
I think AI will be rehabilitated in the eyes of the American public when really wonderful things like cures for cancer come out of it. It will have to prove its usefulness.
I also think that the AI development path suffers from gigantism— the conviction that simply building the database of everything is the way forward. I think smaller, dedicated AI tools that consume less compute, energy, and other resources are the way forward. It is true that you can peel a potato with a 10-foot knife, but there are more efficient ways to do it.
from the people but not for the people
Ben, the real shame is that Google and other major data center builders learned in the early 2000s how to reduce power use and operate more efficiently. Many of today’s AI data centers are going up so quickly that they aren’t always taking full advantage of solar, thermal, and other proven approaches developed over the past 20 years.
And when local or state governments eliminate or sharply reduce taxation, you can see where some of the public pushback comes from.
While there are many small-scale versions of the rot that we could point to (everything-as-subscriptions, algorithmic marketplaces squeezing both sides of the market, ad models incentivizing ragebait and addiction), I think most of it boils down to the requirement to grow at any cost.
This is not, however, universal. In the old days when Google and Microsoft were racing to see who could build the largest search index for their search engines (a more innocent time), eventually Google realized that there were seriously diminishing returns to a larger index compared to the cost to support it. Today Google's search index is basically hard-capped at a specific size, and "least recently used" sites are evicted, which explains why it's impossible to find websites from the early 2000's through search.
In my time in the datacenter world, I asked some Google VPs around 2018 or so, "At what point will Google stop growing its datacenter fleet?" They were unable to comprehend the idea. It has only gotten worse.
Partly I think that this attitude will only be changed by active antitrust enforcement. Even this morning you hear executives saying things like "the anti-datacenter thing is just a phase," because their stock price depends on them believing it. Breaking up a company is the only way to get through to the MBAs that they need to change.
And on the more grass-roots side, we can't get people off their tech dependence without actively helping them. And that requires us to get out there and talk to people about their tech problems, and then help them fix them.
For example, my daughter's daycare had their website enshittified by their hosting provider, and they were required to pay $100/mo to keep it online (and they had no idea how to update it). I redid their website as a free static site. I talked to a teacher friend and helped them write some auto-hotkey scripts to make grading student work easier. I helped another friend use AI to automate pulling stats from websites that hosted their content into a portal they control. I helped another friend deduplicate and compress decades of data they had managed ad hoc over the years. Another compress the images on their website to reduce bandwidth costs.
In doing this, people see tech as a useful tool that they can harness, and it becomes much easier to separate the technology from the extractive company that leeches on their wallet and attention. The fact that most of my friends in tech aren't doing stuff like this (outside of family members) and the fact that most people have to be prodded a bit to open up about their individual tech problems, I wish I had a simple fix for.
Robots can't create surplus value, but they can sometimes create consumer surplus. Automation makes it easier for people to survive but also makes their lives feel more worthless. Short of "destroy all the capital stock in a WWII style orgy of destruction" I am not sure there actually is a solution.
The tech CEOs and thought leaders have so much contempt for government, voters, and the public at large that they literally never thought they could do anything to limit them in any way. The only PR that matters is PR to investors and other business executive customers. So the "AI will put everyone out of work and usher in a permanent underclass" line they've been slinging for years now made sense as a sales tactic.
Pretty telling that the datacenter backlash in the last year is making Altman and Amodei suddenly backtrack and say "actually maybe all white collar jobs won't be replaced..."
> I’d love to hear your thoughts on how the tech sector can clean up its image and make a positive, populist case for itself.
There needs to be a radical shift in how out industry manages diversity. It's maddening that the people who will be affected most by the current gen-AI scourge have effectively zero representation in industry or academia. The real answer is that we need to burn capitalism and the ideas that bolster it to the ground, but that seems to be even more of a pipedream than my current proposal.
*Are* people being made miserable by AI? How much of the hostility comes from experiencing the technology, and how much from what it has come to represent: tone-deaf founders, their politics, and broader economic insecurity while the industry vacuums up billions in investment?
I mean, really, how often do most people encounter AI in everyday life, and how often do they even recognize it as AI? I reckon good uses of AI often go unnoticed, like good CGI is often invisible. But ok, other than through chatting with me, my parents for instance only ever see AI through the Google summaries at the top of their search, some generated images, and occasionally in a product/tech support chat, and they are probably more representative of the average person than chronically online people like me. I have a hard time believing those encounters alone could explain such hatred.
I don't have data on this, but my guess is that much of the hostility is rooted in resentment toward the people and circumstances surrounding AI's development, amplified and reinforced by news and social media.
> How much of the hostility comes from experiencing the technology, and how much from what it has come to represent: tone-deaf founders, their politics, and broader economic insecurity while the industry vacuums up billions in investment?
I get what you're saying, but isn't it the case that the purpose of a system is what it does? Could we have gen AI without the foundation we describe (as in, does that make historical sense)?
How about: take seriously (or at least read) the social critiques of technological society that date back to the Industrial Revolution and dawn of quantification/optimization?
Can you please provide references. Do you books like Technopoly?
These are some that I like:
- Winner, The Whale and the Reactor
- Collins and Pinch, The Golem (h/t Ben)
- Ellul, The Technological Society
- Lyotard, The Postmodern Condition
- Hacking, The Taming of Chance
- Weizenbaum, Computer Power and Human Reason
- Polanyi, The Tacit Dimension
- Feyerabend, The Tyranny of Science
And several good readings here: https://sites.tufts.edu/models/readings/
I have a mass psychology-based explanation for data center hate. It’s largely unfalsifiable and based on personal anecdotes, but hopefully people find it interesting.
Technological innovation has, since the Industrial Revolution, run counter to the immediate interests of the masses in a market economy. Part of the American ethos, however, was that we Americans take a gung-ho attitude towards the new industries that are opened up by new technology, hoping to jump on the bandwagon and see success. Over the long run, this doesn’t work for the majority of people. This is the observation that created the first progressive era.
However, 20th century attitudes towards communism created an environment in which this observation becomes stifled in the mind of the average American. Complaining about the decline in real wages is fundamentally un-American to the masses, because we believe in the free market. In red states, among the most communism-hating sectors of the population, they watch from a metaphorical cuck chair as Musk develops new rocket technology. They cheer him on despite neither being involved nor directly benefitting. Moreover, they cheer him on despite the fact that the market economy which allocated the capital for these developments has left them behind.
Now let’s consider AI. There is no hope, even theoretical, that someone in the boonies of Texas will become involved in the industry behind AI. It is so powerful and all-encompassing that anyone outside of the industry worries about being replaced and becoming superfluous to those in the industry. Far from the days where Ford required a large number of factory workers to build an empire, for the first time we see a technology that can build empires almost without anyone at all, except for a tiny minority of highly educated elites mostly residing in Silicon Valley and traveling unionized construction workers. Those left out rightfully feel hopeless. They should’ve started feeling that way long ago, but now it is undeniable.
However, a religious belief in the free market is still in their DNA. They will never complain about being replaced publicly, or even to themselves. They will instead funnel this anger into a much more American form of complaint: NIMBYism. No data centers in my neighborhood!
As for what tech people can do, I think we have a much larger issue than a PR issue. We have a structural issue. Capitalism is reaching its natural conclusion as new tools emerge to automate everything. We as a society need to agree on a system that distributes the benefits of an autonomy-dominated world to a population much larger than its market economy can support. And we are utterly fucked if we continue on our current path. These elections are a small step in the right direction, and historically large steps in the direction of socialized economies don’t work well—I’m hoping for many more small steps as we approach the final hour.
all fifty measures is kind of wild, i figured at least one town would take the money
There will be no reboot before a crash.
As to the fix, in my view the whole thing needs to be re-thought and re-implemented with reliability and trust-ability at the core. Starting with the training data + labeling, maybe it could handled in a more open, collaborative, manner; perhaps more like Wikipedia than GTA.
Quit your job