In defence of AI, it really depends on what you're measuring. Exploring a new library is probably not what you should be using it for, or a good benchmark for how effective it is.
There's a lot of drudgery in software engineering (like with any job) that AI is fantastic at alleviating, such as writing endless reams of YAML for Kubernetes or Terraform. Another thing is multi-tasking; I can ask it to chew over a problem while I work on something else, and it'll often have done at least 80% of the figuring-out of the problem by the time I check back. It's also great for helping fix bugs, because 90% of that is often just "tell me where in the codebase $THING happens".
That said, this really only help's Martin's thesis that AI is more of a force-multiplier than something coming for my job, unless I fail to get on the bandwagon, like people in 2010 who never figured out they could google things.
In defence of AI, it really depends on what you're measuring. Exploring a new library is probably not what you should be using it for, or a good benchmark for how effective it is.
There's a lot of drudgery in software engineering (like with any job) that AI is fantastic at alleviating, such as writing endless reams of YAML for Kubernetes or Terraform. Another thing is multi-tasking; I can ask it to chew over a problem while I work on something else, and it'll often have done at least 80% of the figuring-out of the problem by the time I check back. It's also great for helping fix bugs, because 90% of that is often just "tell me where in the codebase $THING happens".
That said, this really only help's Martin's thesis that AI is more of a force-multiplier than something coming for my job, unless I fail to get on the bandwagon, like people in 2010 who never figured out they could google things.
Agree with this. I'm just pleased I'll never have to write a manual regex ever again.