You've talked about a very important topic that has now become critical. The purpose of technology and automation is to tackle the challenges, not create new ones. While engineers and scientists worked on how to address the challenge of information overload, the community of engineers and scientists themselves created the issue of overload of a different sort, be it publication overload or AI models overload.

Your suggestion of hypothesis-based development is a good one and can bring in much needed focus and discipline in the AI community to create models with a purpose in mind, and to avoid redundancy.

Was happy to read such an article. Hope other AI engineers and scientists also agree with this and follow it in their professional lives.