#Ai-Progress
2 notas · última:
Empiricists vs. Extrapolators
The essay argues that AI progress is better understood by extrapolating from scaling laws and first-principles models than by waiting for more data, because empirical caution often underestimates exponential change.
What the hell happened with AGI timelines in 2025?
The episode explains why AGI timelines shortened in early 2025 and then moved back out later in the year, pointing to limits in reasoning generalization, inference scaling, reinforcement learning, and autonomy.