Let’s say that at some point we have AGI or even ASI. I’ve noticed that while many experts believe it could happen within 5 years—and that we might no longer need engineers by then—I still have one question: is there any chance that within those 5 years, both PMs and clients will suddenly become capable of clearly, completely, and logically defining requirements?

Anyone with real engineering experience knows that even with years of design work under your belt, it’s rarely possible to determine the perfect implementation approach in one go. So even if we have ASI 5 years from now, does that mean no one will need to define requirements anymore? I doubt it. Requirements often have flaws, and most of the time it’s the engineers who raise concerns and help users understand when a requirement just isn’t feasible.

Anyone who’s been involved in end-to-end design and development and has around 3–5 years of experience probably gets what I mean. But right now, the market is flooded with fear. Personally, I respect and look forward to the development of AI. As for the fear-mongering, I just treat it like advertising.

Perfect requirement descriptions usually come from collaboration among multiple stakeholders. Even when we implement exactly what the client asks for—knowing full well there may be logical issues or limitations—it often still needs changes once it’s put into real use. Can writing out requirements alone really solve everything?