From Lambda to LLMs: AI Coding and the Future of Programming
summary
Adoption jumps from 72% to 82% in weeks and 25% of new code is written by AI today; tools improve weekly and already cut equivalent engineering cost from roughly $20,000 to $20 for many tasks. Scenarios put AI-written lines at ~90%/60%/40% depending on progress and regulation. Hybrid LLM plus proof-assistant workflows use dependent types (Lean/Coq-style) to produce machine-checkable proofs with benchmarks reaching 90% correctness on hard theorem problems, reshaping curricula toward foundations, proof skills, and system design.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 3:57AI coding tools adoption
- 8:14Social impacts and job security
- 12:57AI coding adoption scenarios
- 17:34Programmer roles changing
- 21:08Redefining expertise
- 26:46System engineering complexity
- 29:59Functional programming future
- 33:11Trusting AI black-box code
- 38:53Correctness proofs and FP
- 41:01Educational shifts
- 42:23Q&A
related talks
26mThe present and future of programming with AI (Keynote)
45mProgrammer’s Apprentice Season 2: Future Directions in AI-assisted Coding
24mHow AI is increasing the surface area for serendipity
22mAdapting to the Future of Work in Tech
44mJoint Keynote Address: Accelerating Software Development: The LLM (R)evolution
26mAI-powered Productivity using Open-Source LLMs
32mCan We Trust AI-Generated Code?
43mAI Development Lifecycle: Learnings of What Changed with LLMs