Automation Isn't Innovation
Business is all about marketing and yelping their products. But technology works differently.
Innovation in technology has historically grown at a superlinear rate. Web technology, for example, is not new, yet it never stagnated after its creation. Instead, it continuously evolved. On the frontend, we saw the emergence of increasingly powerful libraries and frameworks, from Angular to React to Next.js. The same pattern applies to backend systems, tooling, and infrastructure.
In artificial intelligence, the current state of the art models are Transformers. At their core, they predict the next token based on previous tokens. The pipeline is simple in principle: data is collected, a model is trained, and the result is a system that predicts patterns extremely well.
AI, however, does not create innovation. What it excels at is automation. Current architectures are nowhere near AGI. In fact, no one even agrees on what AGI truly is, and these architectures will never reach it. What the industry is doing instead is scaling model size aggressively, which is an inefficient and unsustainable approach rather than a fundamentally new idea.
If you are a software engineer at a company rewriting the same CRUD code repeatedly, then yes, you are at risk of being replaced by AI. I have seen many excellent developers who can truly build systems and write real software, not just solve Leetcode problems.
Big tech companies have massive funding and have produced tremendous innovation over the years. I do not understand why the narrative has shifted to saying that developers are no longer needed or that AI can replace them. AI combined with software engineers should open more doors for innovation, not fewer.
Instead of growth, companies are focused on shrinking. Ironically, software engineers should be the last group to be replaced, because they are the ones who build and integrate automation. Replacing them first is a fundamental mistake.
I am writing this because almost no one talks about these issues openly, but everyone yelping about Claude Code. If Claude is really that good, why did they choose to acquire Bun for one billion dollars instead of hiring senior developers to build it themselves?