Special relativity of superintelligence
Human intelligence’s impact grew as people preserved and combined knowledge through language, teaching, writing, institutions, and technology. Tools extended what people could do, and shared knowledge let later generations improve them. Intelligence and tools formed a snowball—not because brains evolved faster, but because ideas became cumulative and connected.
In fact, we may have become less intelligent over the past few decades because of our growing reliance on powerful tools.
AI may follow a similar trajectory on a different clock. Digital models can be copied, tested, and run in parallel. If AI helps improve software, research, or hardware, it may accelerate its next iteration. Cycles that take generations for humans could happen faster digitally.
The bold claim is that the rapid progress of superintelligence is an illusion. Rather, the “brain” of SI was born in 2022 and reached adolescence around 2025. It hasn’t become much more intelligent since then, but its ability to leverage external tools and controls has snowballed.
Enjoy Reading This Article?
Here are some more articles you might like to read next:
- Training an agent that outgrows you is a false proposition
- AI in mathematics may be less like a calculator and more like a tractor
- What should students learn in the AI era?
- When AI is a better prover, perhaps our most important job is deciding what is worth proving.
- The AI Safety Paradox
- Double descent, grokking, enlightenment, wisdom, aging, ...
- Structure of Intelligence
- Makeshift Operation
- Genome length in seconds predicts lifespan?
- When machine intelligence surpasses human brain, human, not robots, is the tools in the age of machines
giscus comments misconfigured
Please follow instructions at http://giscus.app and update your giscus configuration.
Missing required keys:
repo_id, category_id.