I’ve noticed as I’ve been working more to integrate AI into my workflow that I’ve tended to be coding in Access VBA a lot less. It feels like maybe we are heading into a new era in programming akin to when higher level programming languages that could be assembled and compiled began to take the place of assembler level programming where you were working more directly with the actual hardware of the system.

It seems that now we have a level above the higher level programming languages with AI models taking our natural language and writing code and being able to make design recommendations and even design changes you can import back into your Access database. Learning how to do this is transformative and in my opinion the way of the future of coding and development.

But, just like the promise of Cobol, a programming language touted as being so easy that managers would be able to create their own reports, the promise of AI that anyone will be able to generate and maintain any app, whether on the web or in Access or anything else, falls way short. To me, AI models are simply a new level of setting up instructions, guardrails, techniques, etc., to make sure it can actually do the job. We’ll still need experts (like us) to make sure things keep running this way and it can still be a full time job. It can be faster, and usually is at first, but AI writing bad unmaintainable code is still a big problem and it costs more and more to maintain code bases like that because it’s just as hard for AI to add new features to a fragile code base as it is for anyone, it can just screw it up faster.

Still, the AI revolution is not going away, it’s best to learn how to use these new tools at what they are best at and not to use them at what they are not so good at. Knowing and learning the difference is one of our newest challenges.