AI capabilities and risks
AI is currently overvalued and forced into applications without clear ROI, leading to issues with accuracy due to poor training data and potential dangers from general AI models.
- AI is overvalued economically and is currently in a 'fad cycle' where companies are forcing it into everything to find a good return on investment.
- AI often generates responses from unreliable sources like Reddit, leading to inaccurate or 'dumb' information due to a lack of source moderation or weighting.
- General AI is considered a 'stupid goal' compared to multiple specific AI models, as specific AIs are more effective and less likely to cause widespread issues.
- There are concerns about AI generating 'slop' code when trained on the internet due to the prevalence of bad code online and the risk of critical errors in applications like military analysis.


