James King

I have been toying with AI lately and I feel that I am learning a lot, but I have also realised that “learning AI” is not a goal.

So far in using AI in the real world I have summarised documents; created superhero ideas and stories; and analysed some data.

But really you could say I was just tinkering and that what I have built has been a series of one-off experiments.

It has been fun and I have gained some time savings, some new insights and some other value.

But I would not say that I have learned AI yet.

I will continue to play with the pieces of AI that I have and to do some courses, but I think I need to know more before I can say that I have learned AI.

My next step should, I think, be to integrate AI into my “normal work” and to build some tools, patterns or applications that I can reuse and makes sustainable.

That would include getting really good at prompt engineering and at evaluating/testing the integrity of what I get back

I don’t think that is enough though. Even if I can say that I can “use AI”, I am only moving from tinkerer to home handyman.

As a tinkerer I build things to use myself and can thus put up with adhoc, flaky structures in the way I can build temporary structures in the backyard. But a home handyman would need to build things properly so that they are used by others safely and they stay up beyond the next storm.

As a home handyman I should be able to tweak things, make simple repairs to things build by others and create things that are useful.

To do that I suppose I need a little more understanding of programming, but only in the way that a handyman understands the tools he has and the limits of his skill.

A that point I would still not rely on myself to build a full corporate solution to AI. For that I will collaborate with others like myself but also expect more from those who call themselves AI experts.

Just as I would expect a tradesperson, artisan or craftsman to bring more to the table than a home handyman would bring, I expect the “expert” to understand the appropriateness of solutions, the safety and sustainability requirements of solutions and many other things.

If I want a kennel for my dog (I don’t have one) then I might build it myself, but if I want an extension to my house I will hire a real carpenter, a real builder and so forth.

But if I got together with a bunch of people who are at my level we could build a corporate AI solution – just as people could go from a kennel to a shack to a shanty town.

But corporate solutions should not look like tinkering sporadically and they should not look like a shanty town of applications.

So part of my learning journey at the moment is to understand what the difference is between “tinkerer”; “home handyman” and artisan in the world of AI.

If someone is an expert then I expect them to understand the security implications, the maintenance costs, the required change management and the standards needed to create something of real quality.

There is so much out there about “AI will revolutionise the world” that I was suspicious that there was too much hype. But now I know a little more, I think it is more like going to the pub and asking people if they can build something at your house.

There are a lot of blowhards who can start a project, but not so many who can finish a project to the quality you would expect from a carpenter or software engineer.

So know I know – hire an engineer for engineering, but feel free to tinker and try your hand at building things for yourself. But also learn more about what real AI is.

So I am learning just a little governance, just a little security and just a little about hallucinations. My aim is to get good enough to be a handyman AND be good at collaborating with craftsman and “experts” who can help me in the areas that I will not be mastering soon.

How is your AI journey going?