Across the globe is a push for someone, anyone, everyone to start using AI for something, anything, everything. Companies are educating all employees in how to build their own applications and solutions with AI. Performance goals around AI are mandatory. It is as if there is an expectation that with AI anyone can and will build solutions that will magically improve their performance.
This is not a new thing actually, it is a well established concept called citizen development. For more decades citizen development has been a possibility and there have been a half-hearted push for it to allow business people to build their own applications. This has been on the Power Platform lately but if we go back you will find many entertaining and not so entertaining business critical applications built on previous state of the art citizen developer platforms like Excell and Access.
The outcomes of citizen development are well known: a proliferation in ungoverned shadow IT that brings in enormous amounts of technical debt, security risks and undocumented dependencies to vendors and solutions that will only ever really become apparent when things start to break. And then we haven’t even started to talk about all their non-functional deficiencies such as:
- Back-up and recovery (what happens when somebody accidentally deletes the folder for Bob’s homegrown Bank Account Management system so now we can’t pay vendors?)
- Scalability (what happens when all of procurement starts throwing invoices at Jenifer’s auto bookkeeper app)
- Adaptability (what happens when Kevin leaves the company?)
- Security (how could anyone know that Sandra’s public HR support bot could be used for getting private data on employees?)
Everybody knows that citizen development is the gift that keeps on giving, yet curiously there is a worldwide industry agnostic drive for self-extinction through random, mostly overlapping AI solution built by everyone and their dog to solve top of mind problems like the autonomous pigeon deterrent, the sourdough evaluator, the Slack status faker, or my favorite: the “Which Meeting Could Have Been an Email?” Evaluator.
Basically, the AI revolution has opened the floodgates for a new wave of amateur software resulting in AI sprawl. There is nothing wrong with democratizing technology, but it should be clear what gets lost when you turn over the reins to the amateur, ie. the things you expect from professionals. As I have argued earlier : “The software is not the product. The product is the knowledge, organisation and processes that make it possible to produce the software and you can’t vibe code that.”
Even if AI can answer all questions, the bigger part of being a professional is to know what questions to ask. This points to a much deeper misperception of what AI is and can do, because intelligence is seen purely as problem solving, but the other, and more important half of intelligence, is problem finding (as I have argued at length in this article with much philosophical fluff). The difference between the amateur and the professional in any domain is not so much their skills as the ability to ask all the right questions. If you are not able to find the right questions, it doesn’t matter that AI can find all the answers for you. If you as a citizen developer don’t think to ask questions like:
- “What happens if the application is deleted?” (you fail to design for back up and recovery)
- “How do I ensure that the solution continues to be useful when the usage increases?”( you fail to design for scalability) or
- “how do I make sure that it is possible to continue to develop the solution?”( you fail to design for adaptability)
- “How do we make sure that only authorized people access secret data?” (You fail to design for security)
AI is a transformative tool that empowers people to do many interesting things, but just giving the tools and encourage everyone to go build is a well documented recipe for disaster and AI doesn’t change any of that. If we gave people the tools to build their own vehicles would we expect something good to come out of it or traffic Armageddon. It is time to align expectations of what happens in a new world dominated by AImateurs and brace for impact or to maybe rethink whether the professionals actually still have something interesting to offer and build an operating model that can funnel the creativity of everyone and fuse it with enterprise grade quality.