On AI Skills…

Last week, I listened to Shafqat Islam, Optimizely’s CMO, talk about how the Opti marketing team created some of our customer case studies.

What we did was interview a customer for 30 minutes about their project. Then we gave Opal – our built-in AI tool – the transcript along with all the other data we had on the customer. What came out was a solid draft, which was handed off to a human for final polish and publication.

This got me thinking about the skillset for using Generative AI (or, as I’ve started calling the entire toolset: “Editorial AI”).

When programming languages were invented, people were amazed at the magic behind them. But in reality, the languages took a bunch of math and “transmuted” that skill and workload into another one: programming. The new process was more powerful, but it had a learning curve. Mastering the new skill put you in a better position, but it wasn’t free.

Look at no-code UI building tools. They eliminate a lot of coding, but they’ve just transmuted that skill. Now you have to understand the tool itself, and – more importantly – since these tools have opened up UI development to a whole new group of people, it’s forced these people to develop design skills (when you don’t use a designer, you become the designer…). If you invest the time to learn that skill, you’ll be in a better position than before, but it’s not free.

When we domesticated horses, we needed to learn how to ride them and deal with all the externalities they created. Where do you put them? What do you feed them?

Same thing with cars. Same thing with airplanes.

The story of technology is a story of transmutation – we eliminate some process or skill, but in doing that we create a need for some other skill and/or transfer a workload somewhere else. If the new skill and the “somewhere else” are easier and more advantageous, then it sticks. But now we all need to learn something new.

This is where we are with Editorial AI.

Let’s go back to the process of writing case studies that Shaf detailed –

Sure, Opal got us to a nice first draft, but that wasn’t free. Someone still needed to plan that process. They needed to know what Opal was capable of. They needed to coordinate and execute the interview. They needed to guide the customer into telling their story. They needed to know what information Opal needed to craft something great.

These things are quantifiable skills.

“Prompt engineering” was a simplistic early attempt to describe this transmutation. That label was our way of saying, “AI has transmuted a lot of skills into this new one. You should learn it.”

Optimizely CEO Alexander Atzberger has said, “AI won’t take your job. Someone who knows how to use AI will take your job.”

The benefits of Editorial AI are not free. Like everything that came before, Editorial AI has just transmuted a skillset and a workload. Now there are new skills to learn and new processes to design.

The result will be better than before, but it’s never free.