Speakers: Cody Frenzel & Laurie Lay
See the DevNexus live blog table of contents for more posts
General
- Skeptics are useful
- Don’t shut how haters down
AI Adoption Metrics
- DORA – includes how often deploy and lead time to deploy changes
- Developer time savings
- PR throughput (instead of % of generated code)
- Utilization, impact, cost
Other notes
- Don’t mandate AI
- Measure what matters
- AI gains depend on foundation. Technical excellence matters. ex: testability, code reviews, quality gates
- AI will write imperfect code; just like humans. Guzrd rzils prevent it from getting to prod.
- Culture still matters more than tools
AI Literacy
- Tool churn is normal for a new ecosystem. Just like JavaScript in the early days.
- Maintain fundamentals. ex: code review, systems thinking
- We learn through repetition, If we outsource that repetition we don’t learn. Juniors need to write by hand to gain intuition on how to program.
- For seniors, can make instincts weaker, dull senses, lose detecting problems like scale. Need to have non AI periods. Don’t want to be able to assemble but not maintain
- AI use involves self awareness
Things to hate include
- AI slop
- Bad ideas
- Too many tools
- Prompting is a skill
- AI makes you week
My take
The Women in Tech lunch ran late and then I was talking to someone so I was 20 minutes late to this session. It was easy to jump into following from when I walked in though. I like the format of having the 10 things to hate and highlighting them in small groups to talk about concepts