Losing jobs to ChatGPT? No way.

May 11th, 2023. Davorin A. Dumančić


I’ve recently had my share of playing around with ChatGPT. It is a big hype now. It should be. It gives impressive results. But it also concerns a lot of people about loosing jobs. My five cents on that…

We are a development company and have a list of unfinished, low-priority, development requests in backlog. Most of them are not customer related and as there is always a shortage of developers, they tend to be put aside. There are a lot of news on ChatGPT code generation and I’ve taken one of our simple request to try it.

For one of our products we have a health-check services that detects some potential inner-network problems. Can happen once in a quarter. Not a big problem. Restarting EC2 instance in a cluster fixes everything. So I’ve asked ChatGPT to generate AWS Lamba service to: call our health-check service, parse results and if it indicates problems, restart the instance and send notification to administrators via SQS that incident happened. It gave code with some explanations, I’ve just corrected the detection of error code in results as I haven’t specified how to do it. Tried it. Works. It was in Python (don’t like Python; understatement). I guess it should have told him to use proper language. Newbie. But job done!

So it generates code and probably solves some general problems. Does that mean we are going to lose jobs? In ICT business we almost always lack personnel to carry-out all the requests from customers. Getting a new tool to fill this gap? It is great. Maybe we won’t need some junior positions anymore but we can and we will adjust to that. One could draw parallel to any advancement in IT development. From assembler, to C, to C++, to modern object oriented languages, to code generators, to business tools. It always meant that some things would be easier to do and it will take less time and less personnel to do it. No jobs lost. Jobs just shifted to new levels.

In modern IT projects a lot, if not the most, of the resources are spent in bridging the gap between customer non-technical requirements/wishes and technical development team. Most of the unsuccessful projects fail to bridge that gap. And there are a lot of unsuccessful projects. Common statistics, give that up to 85% of IT projects fail. How is this related to ChatGPT or similar technologies? It can relieve us from general technical tasks as mentioned before and enable us to spent more work on resolving problems and requirements with our customers. Can it replace us with that too? No, at least not for now. Every new level of technology that would simplify things for end-users always needed some kind of more technical person to use it and implement business requirements.

p.s. Some prominent persons, like Elon Musk and Steve Wozniak, are calling to stop AI development for six months to see how all this impacts our daily lives and society. Will anything be resolved in six-months? Any conversations on these issues are welcomed by me. Especially, if it involves giving automated systems real execution power (weapons!) but this is not related only to AI. These are the problems that are at least several decades old. Requesting to stop something, that in reality will not be stopped, could just make people feel even more troubled with AI.

#chatgpt #ai