Leveraging AI 227 | AI is now outperforming top human experts in coding & real-world tasks 🤯 Sam Altman's favorite ChatGPT feature just dropped, $100Bs pouring into AI infra, and the economy may never be the same — key AI news for the week ending on September 26, 2025
50 snips
Sep 27, 2025 AI is now outperforming elite human coders, shaking up the tech landscape. OpenAI and Google DeepMind have won top coding competitions using public models. A staggering 42% of companies are adopting AI agents, while predictions warn that many could vanish without adaptation. Major investments in AI infrastructure are ramping up, with NVIDIA backing OpenAI’s ambitions. Plus, the launch of ChatGPT Pulse promises a personalized touch, and debates continue around AI's economic impacts and the necessity for substantial revenue to sustain growth.
AI Snips
Chapters
Transcript
Episode notes
AI Beats Top Coders
- Off-the-shelf models (GPT-5, Gemini 2.5) beat elite human teams in the ICPC coding contest.
- This shows AI now outperforms top humans in well-defined technical tasks and will spread to other fields over time.
AI Performance On Real Work Tasks
- OpenAI's GDPVal benchmark evaluated 1,320 real work tasks across 44 occupations and nine GDP sectors.
- GPT-5 matched or surpassed human experts in ~41% of tasks and models run ~100x faster and cheaper for the AI portion.
Act Now On Reskilling And Oversight
- Prepare for rapid AI adoption and the need for human oversight today.
- Invest now in reskilling and governance because oversight requirements will decline but not disappear soon.