Yesterday, I found an interesting video about the reality of using AI in software teams. The title is "The Hidden Cost of AI Coding That's Destroying Engineering Teams."

 

💡 We know when we create tech debt. We know how to deal with it (it's, in fact, hard!). But AI adds a new type of debt - comprehension debt. Comprehension debt is a gap between the code you have in the software and the code you actually understand. Comprehension debt grows invisibly.

💡 Paradox of productivity: junior engineers generate more code with AI, but senior engineers have up to 19% productivity drop while reviewing AI-generated PRs.

💡 Junior engineers do not build mental models. They borrow mental models from AI. 

💡 AI is an amplifier. It amplifies good processes and engineers. But it also amplifies bad practices. 

💡 The main artifact of work now is not code. It is a specification.

💡 The market needs more expert of veritication as more code is produced. 

So, let's use AI responsibly and effectively.