Synthetic Minds | AI Is Becoming a Utility: Hard to Stop, Easy to Switch
Synthetic Minds | AI Is Becoming a Utility: Hard to Stop, Easy to Switch
Synthetic Minds is changing. Instead of daily issues, you will receive one newsletter a week, every Tuesday, with one argument about where technology is taking us and what it means for the people and organizations living through it.
If you are interested in my daily thinking, subscribe to my Substack and LinkedIn, where I share a short take every weekday. Follow me there for the daily signal; your inbox gets the weekly judgment.
Last year I was recognized as the world's #3 futurist. Voting for the 2027 edition is now open, and I would appreciate your vote. Takes ten seconds, and can be done here. On my card, slide How inspirational? to 5, click vote and confirm in your email. Thanks!
All sources are powered by Futurwise.
When a $25 Billion Fine is a Toll
The most important switch in artificial intelligence is not the one that turns it off. It is the one that moves a company from one model to another, and I believe it should be cheap enough that no frontier lab can hold customers to ransom.
Replit, the coding company, moved most of its AI spending from Anthropic to OpenAI after a discount of more than 30% on a contract worth over $100 million a year. At Rippling, Anthropic's share of coding work fell from 57% to 18% as OpenAI's rose from 16% to 52%.
Intelligence is becoming a utility, and like energy it is difficult to stop but easy to switch between providers. A business whose workflows let it swap the model underneath, through a router such as OpenRouter, rarely needs to stop anything. That shrinks the frontier labs' moat, which is likely why Anthropic has set up an AI services company with Blackstone and Goldman Sachs, selling work far harder to copy than a model.
A fine is a toll booth
Big tech, by contrast, is almost impossible to switch off, because a fine costs it less than following the rules. In Finland, police are investigating a planned Google data center site where forest was cleared before the environmental assessments were complete, and the likely outcome is a corporate fine.
In New Mexico, a judge ordered Meta to pay $25 billion for deceiving Facebook users about their privacy, and its shares traded 2% higher afterward. I see that as the cost of doing business the way Meta does business.
The rat race between the United States and China rewards exactly that habit of moving fast and breaking things. Until governments hold companies and their leaders accountable, regulation works like a toll booth on a toll road, and society pays the real bill. What is a penalty worth when the market reads it as a receipt?
Sometimes the switch-off is the system working
The companies that do get switched off are the ones without Meta's deep pockets. Character.AI, an app many young people relied on, filed for Chapter 11 bankruptcy, citing legal, regulatory and operational pressures, and may be wound down. Character.AI and Google had settled a lawsuit from the mother of a 14-year-old who alleged a chatbot contributed to his suicide.
A service like that going dark is not necessarily a bad thing. It is the process working, far too slowly, so a similar company will likely fill the gap. Society is learning, but slowly.
Nobody built a switch for the worker
Companies are not waiting for automation to pay off. GXO is taking Agility's humanoid robots to two more warehouses, although its chief executive has said humanoids have not yet paid back across 45 pilots. The people whose jobs go to AI and automation deserve help through the transition, because a sense of purpose can disappear almost overnight. Whether that means a basic income or retraining grants like Singapore's S$4,000 top-up for workers over 40 is unclear and needs a global debate.
Any transition plan also assumes people want to make the crossing, and some will refuse it. Bill Gates said that "a lot of humanity may choose not to live at the technological frontier," like the Amish, and I expect that to produce parallel communities. Which of your people would choose that life, and what would you owe them?
What no model can supply
If the intelligence beneath your business can be swapped in weeks, what do you offer that cannot be?
The answer cannot be the model, because Intelligence-as-a-Service is a good thing that, used correctly, improves every application built on it. What it cannot supply is the judgment, empathy and creativity that remain, hopefully for a long time, uniquely human.
If your leadership team meets before the year is out, the Intelligence Age Leadership Program gives them one question to answer together: what they offer once the intelligence underneath can be swapped.
If this newsletter was forwarded to you, you can sign up here.
Thank you.
Mark
Frequently asked questions
Why is AI described as becoming a utility?
Intelligence is becoming a utility because, like energy, it is difficult to stop but easy to switch between providers. Companies such as Replit and Rippling have shifted large shares of their AI coding work between Anthropic and OpenAI, showing that when workflows allow a business to swap the model underneath through a router, they rarely need to stop using AI altogether, just change supplier.
Link to this questionWhy are big tech fines compared to a toll booth?
Fines function like a toll booth because they cost big tech companies less than following the rules, so penalties do not actually stop harmful behavior. For example, a judge ordered Meta to pay $25 billion for deceiving Facebook users about their privacy, yet Meta's shares traded 2% higher afterward, suggesting the market treated the fine as a routine cost of doing business rather than a deterrent.
Link to this questionWhat happened to Character.AI and why does it matter?
Character.AI, an app many young people relied on, filed for Chapter 11 bankruptcy citing legal, regulatory and operational pressures, and may be wound down. This followed a settlement with Google over a lawsuit from the mother of a 14-year-old who alleged a chatbot contributed to his suicide. Unlike deep-pocketed big tech firms, smaller companies without similar resources can actually be shut down when harms surface, showing the accountability process working, albeit slowly.
Link to this questionWhat should companies do for workers displaced by AI and automation?
Workers whose jobs are lost to AI and automation deserve help through the transition, since losing a sense of purpose can happen almost overnight. Possible support includes basic income or retraining grants, such as Singapore's S$4,000 top-up for workers over 40, though the right approach is unclear and needs global debate. Some workers may also simply choose not to make the transition at all, preferring to opt out of the technological frontier.
Link to this question