> The Reasoning Revolution — o1_
Chain-of-thought reasoning. AIME: 74-93%.
> DEEP DIVE_
The year 2024 marked the emergence of a fundamentally new paradigm in artificial intelligence: reasoning models that could think step by step before answering. In September, OpenAI released o1, a model trained not just to generate text but to engage in extended chains of thought, breaking complex problems into sub-steps, considering multiple approaches, checking its own work, and sometimes spending minutes "thinking" before producing an answer. The analogy to Daniel Kahneman's dual-process theory was irresistible: if previous language models operated on fast, intuitive "System 1" thinking, o1 represented the emergence of slow, deliberate "System 2" reasoning.
The benchmark results were extraordinary. On the American Invitational Mathematics Examination (AIME), o1's performance jumped from GPT-4's roughly 12% to between 74% and 93%, depending on the variant. On GPQA Diamond, a benchmark of PhD-level science questions curated to be difficult even for domain experts, o1 exceeded human expert performance. These were not incremental improvements; they represented a qualitative shift in what AI systems could do. Problems that had seemed beyond the reach of language models, multi-step mathematical proofs, complex scientific reasoning, intricate logical puzzles, were suddenly solvable.
The year also brought unprecedented institutional recognition for AI research. In October 2024, the Nobel Prize in Physics was awarded to Geoffrey Hinton and John Hopfield for their foundational work on artificial neural networks. The Nobel Prize in Chemistry went to Demis Hassabis and John Jumper of DeepMind for AlphaFold, shared with David Baker for computational protein design. Two Nobel Prizes in a single year for AI-related work was without precedent, signaling that the scientific establishment had fully embraced artificial intelligence as one of the most transformative forces in modern science.
On the regulatory and economic fronts, the landscape shifted dramatically. The European Union's AI Act entered into force, establishing the world's first comprehensive legal framework for artificial intelligence, with strict requirements for high-risk systems and outright bans on certain applications like social scoring. Meanwhile, the economic stakes reached new heights: NVIDIA's market capitalization surpassed $3 trillion, briefly making it the most valuable company on Earth, as demand for AI training chips far outstripped supply. The combination of reasoning breakthroughs, Nobel recognition, sweeping regulation, and trillion-dollar valuations made 2024 the year when artificial intelligence became the central organizing force of the global technology economy.