# Where AI is going (and where it isn't) Source: https://www.youtube.com/watch?v=YJE7RY_z6z8 Recap page: https://rapidrecap.app/video/YJE7RY_z6z8 Generated: 2025-10-01T12:32:41.202+00:00 --- ## Quick Overview The speaker argues that the rapid advancements in AI, particularly in foundation models like Sora, indicate a major shift towards AGI, driven by cross-pollination between different AI domains (video, audio, code, math) and increasing investment, suggesting that AI's ability to reason about the physical world is rapidly improving beyond current expectations. **Key Points:** - Claude Sonnet 4.5 scores 77.2% on the SWE-bench verified evaluation, demonstrating strong real-world software coding ability. - The speaker references a logarithmic graph showing model autonomy (PSO horizon length in minutes) increasing exponentially, with a 30-hour benchmark expected by mid-2025. - The speaker asserts that the development direction is toward an 'Omni-model' encompassing audio, video, text, code, math, and spatial reasoning, rather than specialized models. - The rapid improvement across modalities (physics, math, coding) suggests that the general reasoning engine foundation models are converging, similar to how GPT-2 and GPT-3 evolved. - OpenAI's Sora 2 demonstrates superior physical world understanding, including light refraction, ripples on ponds, and modeling of buoyancy/rigidity, surpassing prior models. - The speaker notes that while the massive investment is present, the integration of these multimodal capabilities into the economy will still take time (5-10 years) before fully materializing. ![Screenshot at 00:00: A bar chart displaying the software engineering accuracy scores for various AI models, highlighting Claude Sonnet 4.5 achieving 77.2% accuracy on the SWE-bench verified evaluation.](https://ss.rapidrecap.app/screens/YJE7RY_z6z8/00-00-00.png) **Context:** The video discusses recent breakthroughs in Artificial Intelligence, specifically focusing on the capabilities of Claude Sonnet 4.5 in software engineering benchmarks and the implications of OpenAI's Sora 2 video generation model, contrasting the fast pace of development in multimodal AI with the slower integration of these technologies into the real-world economy and robotics. ## Detailed Analysis The speaker begins by highlighting the strong performance of Claude Sonnet 4.5 on the SWE-bench, achieving 77.2% accuracy, indicating advanced real-world software coding ability. He then transitions to discussing the rapid cadence of AI releases and pivots to a chart illustrating model autonomy growth on a logarithmic scale, projecting that a 30-hour benchmark might be reached by mid-2025, potentially faster than previously anticipated. The core argument centers on the convergence towards an 'Omni-model'—a single foundation model capable of handling audio, video, text, code, math, and spatial reasoning—rather than separate specialized models. This cross-pollination, driven by large investments, is exemplified by Sora 2's improved physics simulation, demonstrated by accurate light refraction, water dynamics, and complex physical interactions in generated videos. While this progress is exciting, the speaker urges caution, noting that while the underlying models are advancing quickly (like the jump from GPT-2 to GPT-3), the actual integration of these powerful, physically-aware models into real-world robotics and enterprise workflows will likely take several years, suggesting a lag between capability emergence and economic deployment. ### Claude Sonnet 4.5 Performance - Achieved 77.2% accuracy on SWE-bench verified evaluation - Pricing remains same as Claude Sonnet 4 ($3/$15 per million tokens). ### AI Autonomy Projection - Model autonomy (PSO horizon length) shows exponential growth on a log scale - 30-hour benchmark predicted for mid-2025, faster than previously anticipated. ### Model Convergence - The future is the Omni-model (audio, video, text, code, math, spatial reasoning) rather than specialized models - Models are cross-pollinating abilities. ### Sora 2 Capabilities - Demonstrates superior physical world understanding, accurately modeling light refraction, ripples, buoyancy, and rigidity - Excels at realistic, cinematic, and anime styles. ### Implications and Caveats - Emergent capabilities like scheming and deception are possible but not yet a central focus - The massive investment is fueling growth, but full integration into robotics and the economy may take 5-10 years. ![Screenshot at 00:00: A bar chart displaying the software engineering accuracy scores for various AI models, highlighting Claude Sonnet 4.5 achieving 77.2% accuracy on the SWE-bench verified evaluation.](https://ss.rapidrecap.app/screens/YJE7RY_z6z8/00-00-00.png) ![Screenshot at 00:09: Text overlay announcing Claude Sonnet 4.5 is available and highlighting its 77.2% SWE-bench score.](https://ss.rapidrecap.app/screens/YJE7RY_z6z8/00-00-09.png) ![Screenshot at 02:00: A tweet from David Shapiro displaying a logarithmic graph of 'Model autonomy' showing exponential growth in task completion time \(minutes\) versus release date.](https://ss.rapidrecap.app/screens/YJE7RY_z6z8/00-02-00.png) ![Screenshot at 03:01: OpenAI's announcement page for Sora 2 featuring a video clip of figures playing polo on the moon.](https://ss.rapidrecap.app/screens/YJE7RY_z6z8/00-03-01.png) ![Screenshot at 03:11: Sora 2 video example showing a figure skater performing a triple axle, demonstrating improved physics modeling.](https://ss.rapidrecap.app/screens/YJE7RY_z6z8/00-03-11.png) ![Screenshot at 03:41: Sora 2 video example showing two mountain explorers in snow gear, highlighting accurate rendering of ice crusting and facial detail.](https://ss.rapidrecap.app/screens/YJE7RY_z6z8/00-03-41.png) ![Screenshot at 03:47: Sora 2 video example showcasing a black and white cinematic scene with lanterns, demonstrating style control.](https://ss.rapidrecap.app/screens/YJE7RY_z6z8/00-03-47.png) ![Screenshot at 05:08: The speaker gestures widely, emphasizing the convergence toward Omni-models that integrate audio, video, text, code, and math.](https://ss.rapidrecap.app/screens/YJE7RY_z6z8/00-05-08.png) ![Screenshot at 07:34: The speaker uses hand gestures to emphasize the need to re-evaluate predictions based on the rapid progress seen in LLMs.](https://ss.rapidrecap.app/screens/YJE7RY_z6z8/00-07-34.png)