ChatGPT is not coming for your programming job - unless you suck at programming.

wired.com/story/chatgpt-coding-software-crisis/

Programming is hard. Or rather, programming well is hard. IRL job I have to do Python work from time to time, so I can only imagine what other find with other codes full time.

It's rather like painting: Anyone can pick up a brush and do a quick doodle, but Rembrandts are far and few between.

It's actually worse than painting: A painting just has to be pleasing to the eye to be passable. A program has to work and do what you want to. And a program of even moderate complexity can be a machine with half a million interoperating components, every one of which exhibits non-linear response.

>FORTRAN was supposed to allow scientists and others to write programs without any support from a programmer. COBOL's English syntax was intended to be so simple that managers could bypass developers entirely. Waterfall-based development was invented to standardize and make routine the development of new software. Object-oriented programming was supposed to be so simple that eventually all computer users could do their own software engineering.

None of that happened, because programming is a fairly specific skill.

>What did happen is that programmers could use these new tools to accomplish more complicated tasks more quickly.

We've introduced more and more complexity to computers in the hopes of making them so simple that they don't need to be programmed at all. Unsurprisingly, throwing complexity at complexity has only made it worse, and we're no closer to letting managers cut out the software engineers.

ChatGPT - or its open-source equivalents, like ArbitraryCamelid-7B7 - could make a difference in certain areas such as feature tests and pen-testing. However, LLMs won't and can't by their nature replace programmers, because they don't understand what they are doing in the first place.

The LLMs, I mean. Often the programmers too, but the distinction is, not always.

We'd require a different, older, and harder form of AI to do that, and right now nobody is even looking in that direction.

Reply to this note

Please Login to reply.

Discussion

No replies yet.