The Croker of AI Adoption

In 1969, the original version of “The Italian Job” starring Sir Michael Caine was released by Paramount pictures to rave reviews. The same movie was remade again by Paramount in 2003, this time starring Mark Wahlberg.

Whether you are a Caine fan or a Wahlberg fan, we all fell in love with the character of Charlie Croker. This charming man gravitates others around him and leads a team on a daring heist.

But what does this have to do with Generative AI Adoption?

Imagine what would have happened if Croker had just heard about “the loot” and decided to take a swing at stealing it. He would have been promptly arrested and the movie would be little more than 30 minutes.

The key to pulling off such a huge heist was ….. planning! As John Beckley said, “People don’t plan to fail, they fail to plan”.

But what does this have to do with AI adoption? Well, everything really.

In 2023 and again in 2024 I am still seeing organisations (government and commercial alike) rushing either rushing into AI adoption with no real plan, or avoiding it completely due to no solid adoption plan.

For those of us who have been around technology for long enough, it seems to be history repeating itself.

Why a mini?

In the movie, the chosen car for the heist was a mini minor. In the original 1969 version, even though Fiat offered to sponsor the film with cars, the producers thought it would be more British to use the mini minor. But why not a Rolls Royce with a huge engine? Why not a small truck for that matter?

The mini was chosen after the plan was worked out and the escape path identified. A small, light, manoeuvrable vehicle was needed to fulfil the job.

In the world of AI, too many projects are kicked off by choosing the tech first. Madness!

This distorted approach is often encouraged by the technology vendors with their “free proof of concept”.

12 step framework

When I work with organisations, I take them through a 12-step framework to AI adoption. Choosing the tech is step 10, not step 1.

The critical step (and it is a big one) is to try and identify as many use cases for AI as possible across the entire organisation, not just the customer experience function. These use cases then need to be ranked or prioritised to allow or better planning.

The most obvious use case is not always the best, and the best use case is not always the most obvious.

For example, let’s consider putting a Generative AI front end on to and existing knowledge repository. This knowledge might be in a database, or in SharePoint, or Confluence, or it might just be an overgrown spreadsheet.

Before Generative AI can serve up this content, somebody needs to go through and classify the data into what is confidential and what is not. You don’t want Generative AI serving up the wrong content to the wrong audience.

Once you have gone through the 9 other steps and you know exactly what you want to do, when you want to do it, how you want to do it, and with what data …. Then you can go and look at tech.

It’s all in the planning

One of the coolest things about Generative AI is it is often very quick to deploy, unlike older technologies were. However, the planning and readiness work is extensive and may take months.

So, next time you find yourself drooling over AI’s potential and itching to put it into practice, pause for a moment. Before ploughing ahead, take a step back and embrace that inner Croker.

Immerse yourself in the nuances of the use case, gather the insights, chart the territory, understand the key players, examine the data. Only then are you truly equipped to wield Generative AI as the powerful instrument it can be.