edited from talent to job
I was super nervous AI would replace me, a programmer. So i spent a long time learning, hosting, running, and coding with models, and man did I learn a lot, and you’re spot on. They’re really cool, but practical applications vs standard ML models are fairly limited. Even the investors are learning that right now, that everything was pure hype and now we’re finding out what companies are actually using AI well.
There are a fair number of “developers” that I think will be displaced.
There was a guy on my team from an offshoring site. He was utterly incompetent and never learned. He produced garbage code that didn’t work. However he managed to stay in for about 4 years, and even then he left on his own terms. He managed to go 4 years and a grand total of 12 lines of code from him made it into any codebase.
Dealing with an LLM was awfully familiar. It reminded me of the constant frustration of management forcing me to try to work with him to make him productive. Excrpt the LLM was at least quick in producing output, and unable to go to management and blame everyone else for their shortcomings.
He’s an extreme case, but in large development organizations, there’s a fair number of mostly useless developers that I think LLM can rationalize away to a management team that otherwise thinks “more people is better and offshoring is good so they most be good developers”.
Also, enhanced code completion where a blatantly obvious input is made less tedious to input.
I’ll give you that one. LLMs in their current state help me write code that otherwise I would be putting off or asking someone else to do. Not because it’s hard but because I’ve done it 1000 times and I find it tedious, and I’d expect an entrylevel/jr to take it with stride. Even right now I’m using it to write some python code that otherwise I just don’t want to write. So, I guess it’s time to uplevel engineers. The bar has been raised, and not for the first time in our careers.
