For some time now I have been struggling with this thought and I believe it is worth sharing at this point in time. For context, my LinkedIn feed is overbloated with posts saying that the job market will collapse due to the increased productivity in different industries, but predominantly in software. As I am actively building software solutions in different kinds of organizations, plus building a company from scratch involving AI in the health tech sector, I believe my vision is worth reading. This is not a technical read and I encourage anyone who is curious about how all this AI thing can turn out to read it with an open mind.
"I do not fear AI will boost the productivity of our economy so much that a great portion of the current jobs will be lost, I fear that it won't." Me, right now.
Jokes aside, this is my thesis and I will give my points. I know I have already made most of you mad, please keep reading. I do not intend to predict the future, just to explain a possible outcome different from what I have been hearing for many years now.
Recently I have been reading quite a lot about product discovery. I have learnt that great teams must distinguish outputs (in software, that would be the code our team ships) from outcomes (the final value the code we ship will have on our users). It is critical to measure our success on outcomes, as that is the way our business will produce real value. I will keep this very short, if interested, read "Continuous Discovery Habits" by Teresa Torres.
If you agree up to here, let's move to the next point. AI and every tool available helps us produce more outputs, lots more. It is much easier nowadays to solve any problem that comes to our minds. So this is great, the world will be fixed, just buy lots of tokens!!
Well, a great and significant part of creating amazing products that will really solve problems and that people will actually pay for, is builders talking with customers to really understand the problem. This is because, as I just said, solving a problem is easy indeed, but the thing is that IT WAS ALWAYS THE EASY PART OF THE JOB. The hard thing is not solving, it is knowing what is worth solving. This last part is what worries me. If we integrate AI in a way that helps us produce an absurd amount of outputs in no time, how can we keep up with the validation that we are on the right path?
Just try N number of experiments and you will see the results!! AI enthusiasts will reply. I do not disagree, but if you have not taken the time for discovery, the number of tries your team will do will keep growing exponentially. This is the road to hell. We have sped up development just to spend more time developing the wrong thing.
I believe that providing a real-life example will shed some light on all this. While building Neurosuite (our platform for neurocognitive tests) we intended to partner with commercials so they would help us sell the product in some markets for a cut of each sale. As most of the team were engineers (including myself), we immediately built a whole solution that would let the only partner we had agreed with at that time have great observability through all the sales made, to be transparent with them about everything we had to pay them, we even made a dashboard for them!! All this complex solution "only" took me a week or so, taking into account that we all work part time after our 9-5 job, that's quite fast for the extent of the solution.
So, what's the problem? We did all this before selling anything, and even before we signed a deal with a single partner. In the end, we dropped the commercial partner as we did not have the same vision and reconsidered the need for any of them. So we deleted ALL the code. Deleting everything was not a bad move, the wrong one was building it in the first place.
This is just a small example to show you that building like this can lead to lots of inefficient work, even if that work is done faster than ever. This is not to say that this is because of AI, of course you can make bad decisions. I'm just saying that with adoption of it, you might feel tempted to skip the discovery and the need to identify the correct features to build. Just build it already!! I really discourage this way of thinking. The best teams will continue to put the most effort into product discovery, and only then will they speed up development thanks to AI.
I might have lost everyone at this point. I hope not.
I have been told many times by people I explain this thought to: "So why the fear anyway? In the worst scenario, we stay the same." The thing is that they are absolutely right, the same we stay. The same after spending BILLIONS on an industry that possibly won't produce the increase in productivity that is expected of it. This is exactly how economic catastrophe is created: big errors in investment. This is, ultimately, my great fear.
If you have read through all this, THANK YOUUUU!!!!! I appreciate any feedback.
