As a software developer I share the anxiety of the current, changing software development landscape with many. It's tempting to handwave it and advise myself to learn from the past instead of being a doomsday prophet but the future is murky, as always.
One direction to go in if you are just starting out and have the knack for it, is to be an AI researcher. The area will get saturated early because a) not many companies will be in a position to develop AI agents b) not many core AI scientists will be required - so, there will be high paying but relatively few openings.
However, these systems will require a lot of data engineering, specially early on. AI systems are completely powered by statistics and will need a constant feed of data even after being built. There will be a competition in finding and integrating niche data sources as businesses try to outcompete each other in different domains. Data might become intellectual property and a potential market in patenting datasets, ensuring security, governance and integration might open up. Data engineering will probably focus on these areas. Standard tools might develop - so "engineering" might devolve into just maintenance but with such competition, I think there will be several protected standards to deal with and engineering will still survive to a degree here.
As for general application development, I'm afraid that it will eventually stagnate. Similar to the state that electronics is in currently, it will just become a matter of soldering standard components and workflows. AI manufacturers will be pushing to standardize workflows/components to make the AI agent's job simpler. So eventually application development might become a lower value job, that just requires good prompt engineering.
Of course, if you need build anything that cannot be done with standard components, a deeply knowledgeable engineer will be needed. Again, these engineers will be highly paid but there will be a niche market for it. Also, building the standard workflows / components that AI can use will also require higher degree of engineering, so engineers will want to get into the large companies that are building them.
In effect, three classes of software developers will emerge - low value app developers, data engineers, and select, highly specialized engineers.
One last thing to add - general compute might all move to cloud. So people might not own computers at all - specially if there are fewer software engineers. If most of the compute moves to cloud, powered by AI, people will just own low powered machines and pay a subscription fee for cloud compute. Sorry apple, not more Macs but more Ipad and perhaps a new ISP department.
Top comments (0)