Reading up on GPT-3

ARC proposes that any surprising mathematical fact has a heuristic explanation, aiming to use this to align future AI systems.

An educator describes how AI's ability to complete homework forced a redesign of assessments toward oral TA check‑ins, video demos, and exams, abandoning low‑stakes written tasks.

Epoch AI reports that the cost to achieve a given level of AI performance has fallen roughly 47% per quarter since 2023, outpacing historic tech price drops.

The discussion explores whether automating AI R&D could trigger recursive self-improvement, potentially yielding years of progress in months, with debate on verifiability, data bottlenecks, and alignment risks.

AI will not cause mass job loss for ordinary people because human bottlenecks and comparative advantage ensure complementarity, and the transition will be gradual.

The combination of large language models with reinforcement learning creates a path for continued improvement, making the argument that AI progress will plateau unlikely.

AI tools like coding assistants increase output expectations and cognitive load, turning efficiency gains into more work rather than less.

AI already possesses judgment in domains like investment banking, ethical dilemmas, and art, despite persistent claims that such skills are uniquely human.