Spending More and Learning Less
AI replaces jobs, costs go down, everybody wins. The board nods.
Even if you play along with the replacement fantasy, the economics do not work out the way anyone expects. Assume the most aggressive version where AI replaces entire teams. Every role is now automated. Costs do not go down. Companies will pay the same or more for the same output. Infrastructure, licensing, integration, oversight, debugging AI mistakes, compliance - it all adds up. The money moves from headcount to vendor contracts and compute bills. When that becomes obvious the narrative will quietly shift. It will not be "we saved money." It will be "well, at least you get from point A to point B faster."
Faster does not mean better. I can get from point A to point B faster on the tube than by walking. But if I am exploring a city as a tourist, the tube is not a value add. I skip the side streets, the unexpected discoveries, and the conversations that happen when you move slowly enough to notice things. The speed removes the value.
Business works the same way. A company that produces code faster has less meaningful time to explore what actually matters to the people it sells to. You are moving through the problem space at high speed without stopping to observe. You are effectively burning money faster while learning nothing in return. And learning is the entire point. Understanding customers, discovering what they need, iterating on real feedback - that is where value gets created.
The assumption is that AI will be everywhere, adoption is inevitable, the only question is how fast. But people might not want AI for everything. They might not want it for some things at all. Nobody knows. The future is not a straight line from today's hype to tomorrow's ubiquity. Markets shift and preferences will change. Backlash happens and regulation arrives. Building an investment thesis on "AI will be everywhere" is the same mistake as "if we build it, they will come." Demand is never guaranteed, no matter how impressive the technology.
The honest framing is less exciting but far more durable. AI augments. It makes certain tasks faster, certain explorations cheaper to try but it does not eliminate the need for human judgment, human taste, or human understanding of what a customer actually wants. A business that treats AI as augmentation uses the speed to run more experiments, not to skip experimentation entirely. A business that treats AI as replacement optimizes for a cost structure that never materializes and burns through capital producing output nobody asked for.
A practical note: speed is only valuable if you are paying attention to where you are going. At ChatBotKit we build for augmentation - rapid prototyping that helps you discover what matters before you commit to building it.