korrents

korrents · The Pragmatic Engineer Podcast

Stop being skeptical about AI for development with Charity Majors

Charity Majors · 1h 26m · youtube.com ↗

26 korrents from this recording

1h
Charity Majors did not write this page.

Every claim below is a statement made in this recording, quoted word for word and linked to the second it was said, so you can hear it rather than take our word for it. The wording comes from the transcript published alongside the recording; the sentence above each quote is our reading of the claim, not their wording.

  1. 0:07:24 · watch on youtube.com↗

    Our reading

    Team output is still what matters, however hard the industry pushes to measure individual output.

    Their wordsI also believe that, you know, there's been this whole push towards individual output, but teams are still what matter. Teams output.
    Interesting
  2. 1 min later

  3. 0:08:01 · watch on youtube.com↗

    Our reading

    Equating better with faster is a very immature description of what better means.

    Their wordsWe can go fast. Let's do same thing faster, you know, just boom. And I've come to feel like that is a very immature description of what better is.
    Interesting
  4. 3 min later

  5. 0:10:31 · watch on youtube.com↗

    Our reading

    The step change in AI coding in 2025 came from the harnesses and the tooling around the models, not from the models themselves.

    Their wordsAnd it wasn't actually the models. It was the harnesses. It was all the tooling.
    Interesting
  6. 2 min later

  7. 0:12:04 · watch on youtube.com↗

    Our reading

    There is no point arguing about whether or when engineers will ship code they have not read; the only useful question is what it would take.

    Their wordsWhat I was saying in that piece though was, I think we were right to be skeptical that time. But now, I see the same thing playing out with would you be willing to ship it some code you didn't read? There's no point in arguing about if it will happen or when it will happen. Talk about what it would take.
    Interesting
  8. 3 min later

  9. 0:14:55 · watch on youtube.com↗

    Our reading

    You can now generate ten thousand variants of a function faster than you could write it once, and that economics pushes software towards replacement rather than repair.

    Their wordsI mean, if you if you think about it, you could generate 10,000 variants of a function faster than you could write it once.
    Interesting
  10. 2 min later

  11. 0:17:03 · watch on youtube.com↗

    Our reading

    Lines of code are the wrong artifact to review; what we should be storing and reviewing is an architecture that generates the code to spec.

    Their wordsLike that just does not seem like the ideal artifact. We should be able to store them somewhere. We should be able to have architecture diagrams that we can review and discuss that generate that code to spec.
    Interesting
  12. 3 min later

  13. 0:20:22 · watch on youtube.com↗

    Our reading

    Software engineers behave as though the world exists in the repository, and it does not: the world is production.

    Their wordsit's always weird to me just how much software engineers really seem to believe that the world exists in the repo. It doesn't. It's production, you know?
    Interesting
  14. 1 min later

  15. 0:21:02 · watch on youtube.com↗

    Our reading

    Production is not what happens after development; production is a stage of development.

    Their wordsProduction is not what happens after development. It is a stage of development.
    Interesting
  16. 1 min later

  17. 0:22:03 · watch on youtube.com↗

    Our reading

    Merging should deploy: you should have to stop the train to keep merged code out of production, not push it to get there.

    Their wordsAs soon as you merge, it should be going out. You like you should have to stop the train to make your code not go into production as soon as you've merged.
    Interesting
  18. 5 min later

  19. 0:26:36 · watch on youtube.com↗

    Our reading

    AI-validated pull requests beat human review at exactly the things humans are bad at, which frees the humans to argue about direction instead.

    Their wordsIt's like they have all the wisdom of their most senior engineers looking at every single diff and that is fantastic. Which means that you don't have to worry about remembering and looking and nitpicking and all the things that we're not good at anyway
    Interesting
  20. 1 min later

  21. 0:28:02 · watch on youtube.com↗

    Our reading

    If you are not going to read the generated code, what you need is conformance testing: proof that it performs within the boundaries of the code it replaced.

    Their wordsOkay, if I'm not going to read this code, how do I know it's going to perform within boundaries of the last code that I generated? That is conformance testing.
    Interesting

  23. 0:28:26 · watch on youtube.com↗

    Our reading

    Trust is an account: whatever you withdraw from it by not reading the code has to be built back up somewhere else.

    Their wordsIf you're debiting from this trust account in the creation of the code, it has to get built up somewhere else.
    Interesting
  24. 4 min later

  25. 0:32:00 · watch on youtube.com↗

    Our reading

    There is no human in the loop: you own the loop, and what comes out of it is your work rather than the model's.

    Their wordsAnother another thing we point out is just there is no human in the loop. You own the loop. The loop is yours. The loop is mine. It would not exist if it was not for me. So I am the owner, right?
    Interesting
  26. 3 min later

  27. 0:34:55 · watch on youtube.com↗

    Our reading

    You may not send anyone something you have not read, and if it would take them longer to read than it took you to make, it is slop.

    Their wordsHere's a baseline. Uh you cannot send anyone something you haven't read. And in fact, if it would take them longer to read it than it took you to make it, it's probably slop.
    Interesting
  28. 2 min later

  29. 0:36:42 · watch on youtube.com↗

    Our reading

    AI can be a shortcut around thinking or a way of thinking more rigorously, and for your core job function only the second one is wanted.

    Their wordsYou can use AI as a shortcut to help you not have to think too much. And you can use AI to help you think more deeply and more rigorously. And both of those use cases have their place. But when it comes to your core job function, we primarily want the second one, right?
    Interesting
  30. 2 min later

  31. 0:38:17 · watch on youtube.com↗

    Our reading

    Neither the AI enthusiasts nor the AI sceptics are making it up; both camps are looking at real trends.

    Their wordsSee, the problem is that neither side is making it up. Like they are seeing really scary trends.
    Interesting
  32. 1 min later

  33. 0:39:28 · watch on youtube.com↗

    Our reading

    Both AI camps believe they are the suppressed minority standing up for the truth, and that shared belief is why neither hears the other.

    Their wordsOne of the things that's ironic though is that both of these sides feel like they are the tiny minority and they're outmanned and they're being suppressed and they are standing up for what is truth and valor in the face of the big AI folks or the big skeptics.
    Interesting
  34. 3 min later

  35. 0:42:30 · watch on youtube.com↗

    Our reading

    An account of an AI win is not worth publishing unless it is coupled with what that win cost.

    Their wordsIt's like we need to hear the wins. We need to hear what's We need to hear about what's possible. We need to hear what's exciting. But you got to couple it with the costs.
    Interesting
  36. 2 min later

  37. 0:44:49 · watch on youtube.com↗

    Our reading

    Software is what AI will be best at, because software and AI are both made of logic and language, which is what lets us bake guardrails in.

    Their wordsSoftware is made of logic and language. AI is made of logic and language. And because of that, we can bake in guardrails.
    Interesting
  38. 3 min later

  39. 0:47:20 · watch on youtube.com↗

    Our reading

    Writing is thinking on paper and there is no shortcut for doing that thinking, so handing the writing to AI hands away the thinking.

    Their wordsAnd I have decided not to sink anymore because writing is thinking on paper. And there's no shortcut for doing that thinking.
    Interesting
  40. 3 min later

  41. 0:50:27 · watch on youtube.com↗

    Our reading

    You cannot really understand the code you write unless you are also the one operating it.

    Their wordsThat was always a bad idea because it's split brain. Half of you are writing the code and the other half are understanding it. I would argue that you can't really understand the code you write unless you're operating it.
    Interesting
  42. 1 min later

  43. 0:51:07 · watch on youtube.com↗

    Our reading

    Twenty years of DevOps had one goal, a single feedback loop from writing code to that code in production, and it failed.

    Their wordsBut I also think that in my mind, 20 years of DevOps was really about one thing. Trying to create one feedback loop that connected people writing code to that code in production. And it failed.
    Interesting
  44. 5 min later

  45. 0:56:00 · watch on youtube.com↗

    Our reading

    Everyone should be using OpenTelemetry, because instrumenting code is now literally faster and easier than not instrumenting it.

    Their wordsAuto instrumentation has gotten so good in recent years. If you're using Open Telemetry and everyone should be using Open Telemetry. All of the common patterns like all of the models are trained on them. So, it is literally faster and easier to build with instrumentation than to than not to.
    Interesting
  46. 13 min later

  47. 1:09:15 · watch on youtube.com↗

    Our reading

    If engineers do not hold themselves to a high standard and build efficiently, somebody else will come along and impose it on them.

    Their wordsIf we don't do it ourselves, meaning hold ourselves to a high standard, build with efficiency, constantly be like trying to get better, if we don't do it ourselves, someone will come and do it to us.
    Interesting
  48. 1 min later

  49. 1:10:10 · watch on youtube.com↗

    Our reading

    Kind leaders usually do beat sociopaths in the same role, but only if they are also good at business.

    Their wordsit's it's not enough to be a good person. I believe that people who are kind and care about people can and and usually do do better than sociopaths in the same roles, but only if they're good at business.
    Interesting
  50. 6 min later

  51. 1:16:34 · watch on youtube.com↗

    Our reading

    We cannot say what a junior engineer is worth because we cannot say what any engineer is worth; it is all vibes.

    Their wordsThe hardest thing about quantifying the value of junior engineers is that we don't know how to quantify the value of any engineer. So it's all vibes.
    Interesting