- Avoiding generative models is rational and responsible due to biases and harms.
- Self-experimentation with chatbots and agents is unreliable due to biases and volatile results.
- Tech and productivity research is often poor and lacks rigorous validation.
- The 'AI' Bubble creates widespread bias, affecting studies, papers, and institutional decisions.
- Half-understanding of generative models leads to reliance on biased peer judgements.
- Generative models harm education, healthcare, creative industries, and tech job pipelines.
- High operational costs and infrastructure demands of AI hinder energy transitions.
- Waiting for the AI bubble to pop is the responsible approach.