OpenAI, ChatGPT and Generative AI

[AmareIsGod]

I found this essay, about the new role of the programmer in the current world of AI, interesting. I think the author is largely right, that the human in the loop changes their role from coder to curator; you are now the idea-generator and assurer of quality. But ... I like coding. I am pretty sure I dislike this change and that I don't really want to work for 20 more years as a curator. Career identity crisis!

https://staltz.com/from-coder-to-curator

I got out at just the right time.

Yes you did. It's a crazy world working in the AI space directly.

[Mori Chu]

How much time have you spent in Cursor with Claude Code or anything similar as it relates to coding with AI tools?

I am currently on parental leave, so I haven't used them for a few months, but before that I was using Claude Code and Cursor both pretty regularly at my work. I mostly disliked using them. Their quality for many tasks was pretty good, but like the linked article says, they changed my role too much for my liking. The tools would just spit out lots and lots of code, and now my whole day was spent carefully reading through it to make sure no horrible bugs were hiding inside. I didn't find it very much fun. But I guess this is the way things are heading. I liked the AI for simple features/fixes and for generating unit tests. For writing big things I mostly didn't like looking through its output. And I really disliked the fact that everybody ELSE on my team would use the AI to spit out a whole feature, then submit it to me to review, often without extensively testing or verifying it themselves; wasted a bunch of my time.

[AmareIsGod]

How much time have you spent in Cursor with Claude Code or anything similar as it relates to coding with AI tools?

I am currently on parental leave, so I haven't used them for a few months, but before that I was using Claude Code and Cursor both pretty regularly at my work. I mostly disliked using them. Their quality for many tasks was pretty good, but like the linked article says, they changed my role too much for my liking. The tools would just spit out lots and lots of code, and now my whole day was spent carefully reading through it to make sure no horrible bugs were hiding inside. I didn't find it very much fun. But I guess this is the way things are heading. I liked the AI for simple features/fixes and for generating unit tests. For writing big things I mostly didn't like looking through its output. And I really disliked the fact that everybody ELSE on my team would use the AI to spit out a whole feature, then submit it to me to review, often without extensively testing or verifying it themselves; wasted a bunch of my time.

Definitely was my experience as well. Very code heavy.

[virtual9mm]

Irresponsible for OpenAI to release a model with hidden reasoning. We're pretty fucked.

[AmareIsGod]

'The People Building AI Earnestly Believe That It Could Kill Us All’: Anthropic Researcher Quits Dramatically

https://gizmodo.com/the-people-building ... 2000808968

[virtual9mm]

Talked to a senior guy at Anthropic and apparently the media is misreporting this. Reasoning was not moved to latent space. Just being shown to users less often.