Comments by "Traveller" (@traveller23e) on "A Life After Layoff" channel.

  1. 11
  2. 9
  3. 6
  4. 5
  5. 4
  6. 3
  7. 3
  8. 2
  9. 2
  10. 2
  11. 1
  12. 1
  13. 1
  14. 1
  15. 1
  16. 1
  17.  @bogdyee  Sure, the industry has all sorts of tools that help such as compilers and profilers, and unit tests are certainly helpful, but the fact is that a compiler is deterministic (assuming a bug-free compiler, mileage may vary). Unit tests and profilers are useful for testing the program. Other kinds of analyzers are good for figuring out potential problems. Not sure what you mean by the delightfully vague "cloud tools", but I assume it has to do with tracking network traffic, or maybe load balancing and stuff? But at any rate none of these are both nondeterministic to a significant degree and affect the end result. Hooking up AI straight to the IDE in a position to generate code seems like a vary risky proposition. Your comparison to a plane's autopilot is not very apt, because the autopilots have very precise operational parameters and have a predetermined set of manouvres they can do (like level change, maintain a specific vertical speed, autoland in conditions that allow it, etc). Outside of this the pilots have to take over, and even when autopilot is engaged you still constantly have a pilot flying. AI has no such limits. Even if you tell people to look over the code before copying and pasting it, what are the chances they will do so? And it's a small (and practical!) step from the dev copying and pasting to someone hooking the AI up to the IDE and getting rid of the middleman. Regarding allowing AI to take important decisions, I suspect it's already happening. And the better and more intelligent AI gets the bigger a problem it will become.
    1
  18. 1
  19. 1
  20. 1