|||
Everybody is excited about AI because it writes code so fast. I get it, I'm excited too. But throughout my career, long before AI, how fast we could write code was never the issue. The problem was that we wrote the wrong code.
Any developer who has been around building products for customers knows this. Getting the right product into a customer's hands, one that solves a meaningful problem, is the hard part. We've all written a lot of beautiful code. In some cases we worked really hard on it for a really long time. Then it got into the customer's hands and it was a dud. It just wasn't meaningful enough for them to adopt it.
That tells you something. The constraint, the bottleneck, was never writing code. The constraint was knowing your customer well enough, deeply understanding and empathizing with their problems, to provide the right solution.
To be fair, this is difficult work. It's laborious and ambiguous, and I think that's exactly why it's so easy to avoid.
Humans want to solve problems we have control over. Empathizing with a customer, really understanding their problem, and then being creative enough to invent a novel solution for it is hard. It's much easier to just decide on a solution that we think will work. We ignore the fact that we haven't done the iterative customer development to know it works, and we spend our time building a beautiful solution (or directing our engineering teams to build it). That we can control. So that's what we fall back on.
Now AI has pushed the cost of creating code even closer to zero. But again, that was never really the problem. Code has always been cheap.
If anything, being able to write more code, and more complex code, faster than ever before only compounds the problem. We can now build the wrong thing faster than we ever could.
However, it does bring a new problem as well. Now, we can try more things than ever. We can think about solutions that are truly novel, that in the past we would be too embarrassed to suggest to our boss or to the marketplace because it's such a big, ambitious and "unrealistic" idea. The reality is that soon - even now - AI makes anything possible... If we are audacious enough to even suggest it.
That doesn't mean we get a pass on understanding if our audacious solution is truly valuable to people. So valuable that they're willing to invest their extremely finite time into it. So we still have a lot of work to do. But I believe that now, or at least very soon, the biggest obstacle to progress will be that we're not audacious enough.
So we need to take a bigger step back from the part we have control over, the code. We need to deeply understand our customers. And then we need to dream bigger than we ever have before.
The code isn't what holds us back. Our ability to dream big enough is.