So far, the best predictor for AI progress was compute scaling

**Credit:** Sevilla, J. *et al.* (2022) ‘Compute Trends Across Three Eras of Machine Learning’. arXiv. Available at: https://doi.org/10.48550/arXiv.2202.05924.

Richard Ngo argues that so far, most forecasters that correctly predicted AI progress based their forecasts on compute scaling. Done in this fashion, people like Hans Moravec predict human-level AI by 2030, Ray Kurzweil by 2045 and Shane Legg by 2028.

Source: Ngo, R. (2023) ‘Visualizing the deep learning revolution’, Medium, 18 January. Available at: https://medium.com/@richardcngo/visualizing-the-deep-learning-revolution-722098eb9c5 (Accessed: 9 February 2023).

Sevilla, J. et al. (2022) ‘Compute Trends Across Three Eras of Machine Learning’. arXiv. Available at: https://doi.org/10.48550/arXiv.2202.05924.

Links:
So far, AI progress is driven by the amount of data and compute during training
AI progress is not mostly driven by progress in hardware efficiency