Don't Put The AI Cart Before The Business Problem Horse

Just as the leading edge of AI adoption by business is starting to show signs of wariness the cries of "what is your AI strategy?" are starting to be deafening.

While it is often rude to answer a question with a question, in this case I have no choice. When I am asked "what should our AI strategy be?" I am forced to reply "in what domain?"

Talking about AI without specific reference to the processes to which it is to be applied is to put cart before the horse. General statements about AI are simply not useful as ways of defining how you will apply AI to your daily operations.

AI is a general-purpose tool. Perhaps the most general-purpose tool humanity has ever devised. It is not meaningful to talk about AI as if it had narrow fields of application.

Consider a more narrowly focused utility, such as cloud computing: it is reasonable to ask what strategic decisions your IT folks have made about cloud computing. Do you need server capacity in such great quantities that cloud computing makes financial sense (ie cheaper to buy than build)? Do you need server capacity in such small quantities that cloud computing makes management sense (ie allows you to avoid hiring server experts)? Are you in the grey area between these poles where you have tricky choices between economy of scale and split focus? Do you already have in-house server capacity or not?

Consider a more widely focused utility, such as electricity: it may not be reasonable to ask what strategic decisions your entire company is making about using electricity. We use electricity for so many things and alternatives are hard to come by. We rarely generate our own electricity. We rarely have horses or steam engines to use instead. "What is your electricity policy?" is a question likely to be met with bafflement. What does that even mean?

If we are honest, this bafflement is probably appropriate to the question "what is your AI policy?" We are often not honest in replying to questions about AI because AI is the new thing we are all supposed to be embracing. Few of us dare to reply "what are you talking about?' or "show me the business case for the business solution to the business problems being solved by AI and I'll tell you our policies on a case-by-case basis."

This group think about AI is worse in the realm of cybersecurity. Surely, since AI is disrupting cyber-crime and turbo charging hacking and other scary things that must imply that there are clear applications for AI on the other side of the fence, to protect us better! Right? Right! So, you first: what are these clear applications? And since we are talking about corporate policy to be shared with investors, clients, customers and partners, please make these pronouncement brief, easy to understand and clear. This kind of pressure leads to the kind of banal, anodyne statements we see on the Internet so much these days. (I had originally thought to amuse myself by having an AI generate such statements but the results were too depressing. Try it if the idea amuses you, it is very easy.)

There is no way to formulate AI policies without careful thought and lots of context. Consider this rather good blog post from MIT Management's Executive Education blog: Developing an Effective AI Strategy. But don't expect a crisp and pithy description here: there are no short cuts and one size only fits one organization (and only then if you are lucky)

Figuring out your various strategies for various applications of AI to various aspects of whatever it is that you do is hard. We can help. Ask us how.