An essay published on planned-obsolescence.org makes a bold claim: AI is dramatically accelerating AI research. The argument is that models are now contributing to the very work that produces better models, creating a feedback loop that shortens research cycles. The piece drew 329 points and 291 comments, making it one of the most debated items of the day.
The core idea is not new, but the essay pushes it further than most. It suggests the speedup is already visible in how quickly research moves from idea to experiment to result. If that is true, the implications reach well beyond the lab, touching on hiring, investment, and how quickly the field could change.
Commenters were sharply divided. Some argued that the evidence for a compounding effect is thin and that AI tools mostly help with routine tasks. Others countered that even modest gains in research throughput add up quickly when applied across an entire field. The debate reflected a wider uncertainty about how much of the recent pace of progress is due to AI itself.
The essay does not settle the question, but it frames it clearly. Whether AI is a genuine accelerant for its own development or simply a useful assistant is one of the most consequential questions in technology right now. The strong reaction suggests many people believe the answer matters a great deal.