Russian Federation
employee
This article analyzes potential development trajectories of advanced artificial intelligence (AI) systems towards 2030 based on the extrapolation of current technological and economic trends. It examines the hypothesis that the scaling of computational resources remains the key driver of progress in AI. Based on a retrospective analysis of trends (2010-2024), forecasts are constructed regarding the required computing power for training, investment volumes, data dynamics, hardware development, and energy consumption. The article posits that, should current trends continue, by 2030 the largest AI models will require 1000 times more computation for training than contemporary models, accompanied by investments amounting to hundreds of billions of dollars and energy consumption at the gigawatt level. The expected capabilities of such systems are further explored, with a focus on the automation of scientific research and development (R&D) in areas such as software engineering, mathematics, molecular biology, and weather forecasting. It is forecasted that AI will become both a highly specialized tool and an assistant-agent, significantly accelerating the "digital" aspects of research. In conclusion, key uncertainties, potential bottlenecks (regulatory, environmental, data-related), and the broader socio-economic implications of the projected AI development are discussed.
artificial intelligence, machine learning, scaling, forecasting, computing, investment, energy consumption, research and development, forecast
1. Bruun, L., & Bo, M. (2025). Bias in Military Artificial Intelligence and Compliance with International Humanitarian Law. SIPRI.
2. Additional Protocol I to the Geneva Conventions (1977).
3. Henckaerts, J.-M., & Doswald-Beck, L. (2005). Customary International Humanitarian Law. Cambridge University Press.
4. Moyes, R. (2019). Target Profiles. Article 36 Discussion Paper.
5. Dvaladze, G. (2023). Equality and Non-Discrimination in Armed Conflict: Humanitarian and Human Rights Law in Practice. Edward Elgar.
6. Kwik, J. (2024). Lawfully Using Autonomous Weapon Technologies. TMC Asser Press.
7. ICRC. (2021). ICRC Position on Autonomous Weapon Systems.
8. Reports and working papers of the CCW Group of Governmental Experts on Lethal Autonomous Weapons Systems (GGE on LAWS).
9. Sutton, Richard. 2019. “The Bitter Lesson”. http://www.incompleteideas.net/IncIdeas/BitterLesson.html
10. Sevilla, Jaime, and Edu Roldán. 2024. Training Compute of Frontier AI Models Grows by 4-5x per Year. Epoch AI. https://epoch.ai/blog/training-compute-of-frontier-ai-models-grows -by-4-5x-per-year.
