Super Data Science: ML & AI Podcast with Jon Krohn 890: The “State of AI” Report 2025
61 snips
May 23, 2025 This week, the highlights of Stanford's AI Index Report reveal groundbreaking advancements in AI technology. Discover which models lead the pack in performance relative to their size, and learn about the scenarios where AI outshines humans, as well as the instances where human ingenuity prevails. The report also underscores the staggering reduction in operational costs and the accelerating adoption of AI across various industries. Tune in for insightful takeaways that paint a vivid picture of the evolving landscape of artificial intelligence.
AI Snips
Chapters
Transcript
Episode notes
Smaller Models Improved Drastically
- Smaller AI models have drastically improved in performance over the last two years.
- We now achieve the same capabilities with models 142 times smaller than before.
AI Inference Costs Plummet
- The cost to run AI models has plummeted, making them much more affordable.
- For example, GPT 3.5 equivalent performance costs fell about 280 times from 2022 to 2024.
AI Agents Match Human Expertise Quickly
- AI agents outperform human experts on complex tasks within short timeframes up to 2 hours.
- Humans still lead on longer tasks, but AI already excels at specific complex tasks like coding.