You should start building with AI now. I’ll give you a long list of caveats, but if you’re too lazy to read them, that’s the conclusion.

Thank me later, now go be lazy somewhere else.

The Wait/Walk Paradox or Wait Calculation

There’s a calculation you can make if you want to travel to space.

Let’s say you want to check out Alpha Centauri, our nearest star, or tri-star really, that’s about 4 light years away.

Let’s also say that you fund a solar sail research project and you figure out how to make it reach 1% of the speed of light.

Great, now it takes 400 years to get to Alpha Centauri.

But wait…

What if you research another 10 years? Perhaps you can find a way to reach 2% of the speed of light.

Now the trip will take 200 years instead, so the new space craft will overtake the one you launch today and reach Alpha Centauri 190 years earlier.

But wait…

What if you research another 10 years? Perhaps you can now reach 4% of the speed of light.

Now the trip takes 100 years.

…meaning that if at any time in the next 300 years, there’s a chance we can reach 4% of the speed of light, there’s no point in leaving now.

And that’s why you should start working with AI now.

Wait, What?

You can apply the same method of thinking to AI to understand when and how to apply it.

The argument would go that there’s no point in starting using AI now because tomorrow will have a new an better model that can do 2x as good a job.

It’s the same argument made for buying a PC. Next month, there will be a faster and better and larger model at the same price, so buying now is a waste of money.

Just wait…

Except nobody does. Because if you wait, you will still have to wait, because the same will be true a month from now.

In fact, it would have been true almost every month since PCs became practically available to everyone in the 80s.

And if you followed your argument, you still would not have bought a PC and you would have lost out on all the productivity from using a slightly inferior model.

And that’s why you should start working with AI now.

Wait… Didn’t you just say…

With AI, new models will come out, tomorrow, next week, in a month… They will be better and faster and half the current price.

And that’s awesome news for those that start today.

As long as you pay close attention to certain things, the fact that there will be a new model that’s better, smarter, and works faster in a month is…

…exactly why we have hired software developers for decades.

Software developers grow and learn and become better, so that they can fix the issues they create today.

Would you avoid hiring someone today because you know they’ll be better in a year or two years?

Well, if you did, you would wait forever, because that will always be true – that if you wait just a bit longer, what you pay for will be a bit better.

Individual AI and LLM models do not get better, but collectively, AI assisted development gets better all the time.

What passes as state of the art today will be a joke in the not so distant future, just like that passed as state of the art in the not so distant past is a joke today.

So if you wait now, arguing that things will get better soon, you will wait forever.

And you will never get your project started, and you will lose out on the fact that those future models can help improve what you build today.