Andrej Karpathy: Software Is Changing (Again)

Y Combinator

Summary

Andrej Karpathy explores the evolution of software development in the age of AI, categorizing it into three distinct paradigms: traditional computer programming (Software 1.0), neural network weights (Software 2.0), and the current era of LLM-based prompting (Software 3.0). He argues that large language models are effectively acting as modern operating systems that are reshaping how we interact with technology. Karpathy highlights the importance of 'partial autonomy', building tools where humans retain control, and suggests that developers should focus on creating interfaces that allow for rapid verification of AI-generated work. Ultimately, he encourages a future where both humans and AI agents can navigate the digital landscape through more machine-readable, English-driven documentation and specialized protocols.

Chapters

  1. 0:01Welcome and Intro
  2. 0:43Software 1.0, 2.0, and 3.0
  3. 6:29LLMs as Utilities and Fabs
  4. 9:09Analogy to Operating Systems
  5. 12:57Technology Diffusion Reversed
  6. 14:48Psychology of LLMs
  7. 16:04Cognitive Deficits and Memory
  8. 18:44Partial Autonomy Apps
  9. 20:11Importance of GUIs
  10. 20:46The Autonomy Slider
  11. 22:20Speeding up Verification
  12. 25:09Education and Agents
  13. 29:35Vibe Coding
  14. 30:56Lessons from Autopilot
  15. 33:33Building for Agents

Topics covered

  • Software Evolution
  • LLMs as Operating Systems

+5 more topics (full content when you follow the channel)

Transcript excerpt

Please welcome former director of AI Tesla Andre Carpathy. [Music] Hello. [Music] Wow, a lot of people here. Hello. Um, okay. Yeah. So I'm excited to be here today to talk to you about software in the era of AI. And I'm told that many of you are students like bachelors, masters, PhD and so on. And you're about to enter the industry. And I think it's actually like an extremely unique and very interesting time to enter the industry right now. And I think fundamentally the reason for that is that um software is changing uh again. And I say again because I actually gave this talk already. Um but the problem is that software keeps changing. So I actually have a lot of material to create new talks and I think it's changing quite fundamentally. I think roughly speaking software has not changed much on such a fundamental level for 70 years. And then it's changed I think about twice quite rapidly…