The Hitchhiker's Guide to AI

As IT leaders, there usually comes a time when we are called to a meeting with the senior executive of the organization to discuss the latest bit of technology that the Sales and Marketing machine, and the tech media think we need.

This usually occurs the week after said senior executive has been to a conference and saw a dazzling presentation extoling the virtues and possibilities of the latest technological wonder. The shining star of the moment is AI, and all its variations from ChatGPT, to generative AI, to workforce reduction using AI, etc.

With the executive's face glowing, he or she says: "this shouldn't be hard, the speaker said [insert product name here] can connect to all our data and tell us what we need to know!" The glow soon fades when you have to explain the reality that the project may be a bit more complicated than that.

Your mind may go back to when you stood in front of that same executive, and had a similar conversation around BA (Business Analytics) and BI (Business Intelligence).

All of the data that was essential to successful BA/BI initiatives, is crucial to leverage the power tools of AI.

Depending on your organization (and my experience tells me that there are more than not), your data ecosystem may be secure, but not optimized to be used by AI, BA/BI.

Without good data in the right format, your gas mileage will vary.

That is why you have been working to clean up your data for years, particularly in the legacy systems. Multiple systems exist that are not integrated and data within those systems are not synchronized.

Sidebar: Nobody, and I mean nobody likes to admit this about their data (their legal department usually has something to say about that), but it is a very real situation for many, if not most organizations. But don't lose hope, it is redeemable, but it takes diligent effort.

The dollar figure for your initiative submitted each year in your strategic plan was much higher than the cost of the software or AI tool the executive was so dazzled by. You have some work to do to bring reality into the conversation.

In my opinion, this front end work is the most crucial phase.

Without good, clean, reliable data sources, then the answers given by AI tools will be of the same quality and reliability of the data.

The second thing to consider is answering the question "Why would you implement AI tools?"

Before you jump into a major AI project, you may just want to lead the project to define the metrics - the performance and success metrics that your organization needs (hint: this is much more work than implementing the system). Once these are defined, then you should find the system that best matches your objectives. .

Let me provide a popular culture reference to help you in your discussion.

In the 1970's, Douglas Adams created a radio program, which was later published as a book called The Hitchhiker's Guide to the Galaxy. There is one much discussed plot line revolving around the number 42 being the answer to the ultimate question of life, the Universe and Everything. Indulge me as I quote a passage from Wikipedia describing the number 42.

"In the radio series and the first novel, a group of hyper-intelligent pan-dimensional beings demand to learn the answer to the ultimate question of life, the universe, and everything from the supercomputer, called Deep Thought, specially built for this purpose. It takes Deep Thought 7 and a half million years to compute and check the answer, which turns out to be 42. Deep Thought points out that the answer seems meaningless because the "beings who instructed it never actually knew what the question was." (source: Wikipedia - phrases from The Hitchhiker's Guide to the Galaxy).

I'm not sure about you, but Deep Thought kind of sounds like many AI projects underway now. Expending extensive time and resources to find the answers, but not absolutely sure what questions need answering.

A side note... If Mr. Adams was writing the series today, Deep Thought may have taken much more than 7 and a half million years to crunch the data - we generate terabytes of new information every day. But I digress.

Years ago, I read a book that contained a phrase that captures this well. I forget the exact name of the book which took leadership principles and reworded them as if they came from Genghis Khan. It was written as a parody and is now long out of print, but this maxim will stick with me forever.

"A Chieftain who asks the wrong questions, gets the wrong answers."

(Gently) remind the executive to remember that before buying that shiny AI tool.