Doctorow predicts that when the AI bubble bursts, small-scale AI models and federated learning might survive and continue to evolve. He also anticipates that the bubble will leave behind a large number of people with skills in statistical analysis and data wrangling, as well as knowledge of AI programming languages. However, he criticizes the lack of discussion about what can be salvaged from the AI bubble among policymakers, who are more focused on AI ethics and safety.
Key takeaways:
- The author, Cory Doctorow, argues that AI is currently in a bubble, with every business plan and advertisement featuring the term, even when the business itself has no AI in it.
- Doctorow differentiates between tech bubbles that leave something behind and those that leave nothing. He suggests that while the dotcom bubble left behind a generation of technologists with valuable skills, the cryptocurrency bubble left very little of value.
- He questions the sustainability of the AI bubble, pointing out that the large models are incredibly expensive to make and run. He also highlights the potential risks and liabilities of AI, particularly in high-stakes applications like self-driving cars and radiology.
- Doctorow concludes by suggesting that when the AI bubble bursts, there may be some valuable residue left behind, such as small-scale AI models and a greater understanding of statistical analysis at scale. However, he also warns that very few people are considering what can be salvaged when the bubble is over.