AI Is Just A Tool

My stepdaughter and her husband came to visit the other day. They had recently gone to the memorial service for a good friend of theirs, who unfortunately passed away too young.

Another friend of theirs, who is a professional photographer, had borrowed photographs of the deceased from family and friends. These photographs, some prints and some digital, covered his entire life from the time he was a baby until more recent ones as a young man.

Through the use of AI, this friend created short clips 2 to 5 seconds long using each photograph as a base. Then, he stitched the videos chronologically into a 5 to 10 minute movie.

He also used AI to generate the soundtrack. One AI generated lyrics that described the deceased’s life, while a different AI generated the music and singing voice to fit them.

My stepdaughter showed me the end result, and I got to say it was the best AI generated video I’ve ever seen.

I can only imagine the reaction of the deceased family and friends, but I’m sure they are all very glad and thankful for that video.

Polarized Views

We live in an age that we take our opinions to the extreme. We can hardly have a conversation about a controversial topic. We tend to emphasize our position against opposing views instead of defending our own position.

Not long ago, I saw a comment from someone hating all things AI on a post about a Kindle cover that was 3D printed and painted with an abstract pattern generated by AI. It’s just a Kindle cover!

Of course, there are fearmongers who like to prophesy about AI causing the end of the world. How the machines will rise and destroy civilization. I feel like siding with Bill Burr, who said, you just have to “unplug that thing”.

On the other hand, you have the AI lovers. These are the people who believe AI is the solution to every problem. They use it for everything claiming unrealistic results. Some of them might be profiting from it; others just drink the Kool-Aid.

For me, the most useful way to think about AI is simpler and much less dramatic.

It’s Just A Tool

The reality is that AI is just a tool. A very powerful tool, but a tool nevertheless. And like with any tool, learning it is the fastest way to demystify it.

It doesn’t take long to find the limitations of AI. As a software craftsman, I find myself iterating through the result of AI-generated code. I try different models to solve the problem and compare results. Little by little I make the necessary tweaks to improve generation, but I have never seen a result that didn’t require several iterations to achieve the desired results.

Even with its problems, I’m more efficient when I use AI. Typing was hardly ever my bottleneck. I spend most of my time, as before, thinking about the problem I’m trying to solve, seeking a flexible design that meets the requirements, refactoring until all the deadweight is removed.

I do the same when I’m using AI. The main difference is that I tell the agent what to do, and then correct the mistakes. I make sure unit and acceptance tests are written. I review them to make sure they are clear and focused on the requirements, and I use that as a measure of initial success.

Then, I iterate refactoring, making sure the code is clean and idiomatic. I only create pull requests for review when I’m happy with the results, when it feels like I wrote the code.

I spend a lot of time discussing with my team about their experience. We learn from each other what works, and, most importantly, what doesn’t work. By learning the tool, we are able to appreciate its power without buying into the buzz.

Sometimes, I use it for fun. I’ve tried chatbots, I play Wordle with one almost every day, and I can tell you they are not good at it. I have to redirect the conversation so that it provides better word guesses. Every new game is a trial and error experience, which proves that they are not as capable as we are told.

I use generative AI to create the signature images for this blog. I’m not an artist, and I don’t make any money from these articles, so I can’t justify paying a designer. I’m sure there are people who would do a much better job, but the free, AI-generated ones I use are good enough for me.

Conclusion

You don’t have to use generative AI, but the technology is here to stay. Some people will use it in horrendous ways. Companies will try to replace people with AI. Jobs will change. And much like any other time in history, we will learn our ways and hopefully make our lives easier.

In the end, the value doesn’t come from the model itself, but from the experience, knowledge, or creativity of the person using it.

Be skeptical, but curious. Learn it. Find ways to apply it. Ditch it when it doesn’t work. The scariest part of AI is when people lower their standards to accommodate the tool.