Re: We need a more sophisticated debate about AI

The issue with our AI debate is that there's not a single "problem" but many inter-dependent issues without a clear system-wide solution.

- Big tech monopolizing the models, data, and hardware.

- Copyright concerns.

- Job security.

- AIs becoming sentient and causing harm for their own ends.

- Corporations intentionally using AI to cause harm for their own ends.

- Feedback loops will flood the internet with content of unknown provenance, which get included in the next model, etc.

- AI hallucinations resulting in widespread persistent errors that cause an epistemological crisis.

- The training set is inherently biased; human knowledge and perspectives not represented in this set could be systematically wiped from public discourse.

We can have meaningful discussions on each of these topics. And I'm sure we all have a level of concern assigned to each (personally, I'm far more worried about an epistemological crisis and corporate abuse than some AI singularity).

But we're seeing these topics interact in real-time to make a system with huge emergent societal properties. Not sure anyone has a handle on the big picture (there is no one driving the bus!) but there's plenty of us sitting in the passenger seats and raising alarm bells about what we see out our respective little windows.

perrygeo, 4 hours ago