The AI Update X: O3

In this AI Update Jhave and Scott discuss O3, the new AI model announced by OpenAI, which excels in high-complexity tasks, demonstrating state-of-the-art capabilities in science, structured reasoning, and problem-solving.

References

Brown, N., & Sandholm, T. (2018). Superhuman AI for heads-up no-limit poker: Libratus beats top professionals. https://www.science.org/doi/10.1126/science.aao1733

Chollet, F. (2019).On the measure of intelligence.https://arxiv.org/abs/1911.01547

DeepSeek-AI et al. (2024). DeepSeek-V3 Technical Report. https://arxiv.org/abs/2412.19437

DeepSeek-AI et al. (2025). DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning. https://arxiv.org/abs/2501.12948

Epoch AI. (2024). FrontierMath: A Benchmark for Evaluating Advanced Mathematical Reasoning. https://epoch.ai/frontiermath

Meta Fundamental AI Research Diplomacy Team (FAIR) et al. (2022). Human-level play in the game of Diplomacy by combining language models with strategic reasoning. https://www.science.org/doi/10.1126/science.ade9097

Rein, D., et al. (2023). GPQA: A Graduate-Level Google-Proof Q&A Benchmark. https://arxiv.org/abs/2311.12022

Shao, Z., et al. (2024). DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models. https://arxiv.org/abs/2402.03300

Snell, C., Lee, J., Xu, K., & Kumar, A. (2024). Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters. https://arxiv.org/abs/2408.03314

Sutton, R. (2019). The Bitter Lesson. Incomplete Ideas. http://www.incompleteideas.net/IncIdeas/BitterLesson.html

Xin, H., et al. (2024). DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Searchhttps://arxiv.org/abs/2408.08152

This work was partially supported by the Research Council of Norway Centres of Excellence, project number 332643, Center for Digital Narrative.