When Sam Altman admits we’re in an AI bubble while simultaneously promising to invest a trillion dollars in data centers, something stranger than usual financial excess is happening.
The cultural bubble includes young people's strategic responses to the experts' predictions that AGI and ASI is around the corner, that the majority of knowledge workers will lose their jobs. It includes Geoffrey Hinton's recommendation to young people to become plumbers rather than computer scientists, writers, or inventors. Yet ultimately the fate of the AI bubble will be decided by the comparative cost of human and machine intelligence.
The current strategy is to provide a free service to destroy the competition and then monetize the monopoly, just like Uber undercut the cost of taxis until they mostly vanished and Uber doubled their prices. In this sense the destructive effects of AI on people's ability to solve problems independently is exactly the transformative effect AI needs to succeed. We could conceive of this as the ultimate battle for human autonomy -- or merely as yet one more corporate rentier attempt to turn us all into subscribers, with the expectation that once they succeed and control the monopoly, the quality of service will inevitably degrade.
What is at stake is not merely whether machine intelligence will fail to provide quality solutions to actual human problems at a cost below that of human intelligence, but whether we as a society will entrust imagining and inventing our future to a small handful of private corporations. Can we even begin to quantify that risk?
To a large extent, the capital invested in Data Centers comes from the enormous extractive profits that the Hyperscalers have made and continue to make. AI as we understand it today would be unimaginable without the concentration of market power at the very top.
The cultural bubble includes young people's strategic responses to the experts' predictions that AGI and ASI is around the corner, that the majority of knowledge workers will lose their jobs. It includes Geoffrey Hinton's recommendation to young people to become plumbers rather than computer scientists, writers, or inventors. Yet ultimately the fate of the AI bubble will be decided by the comparative cost of human and machine intelligence.
The current strategy is to provide a free service to destroy the competition and then monetize the monopoly, just like Uber undercut the cost of taxis until they mostly vanished and Uber doubled their prices. In this sense the destructive effects of AI on people's ability to solve problems independently is exactly the transformative effect AI needs to succeed. We could conceive of this as the ultimate battle for human autonomy -- or merely as yet one more corporate rentier attempt to turn us all into subscribers, with the expectation that once they succeed and control the monopoly, the quality of service will inevitably degrade.
What is at stake is not merely whether machine intelligence will fail to provide quality solutions to actual human problems at a cost below that of human intelligence, but whether we as a society will entrust imagining and inventing our future to a small handful of private corporations. Can we even begin to quantify that risk?
To a large extent, the capital invested in Data Centers comes from the enormous extractive profits that the Hyperscalers have made and continue to make. AI as we understand it today would be unimaginable without the concentration of market power at the very top.
Thanks for writing this, it clarifys a lot. I've been waiting for this part of the series! Your insight on the three bubbles is brilliant.
Thanks for the kind words!