Recently, I've felt that many of my conversations have been circling around AI. Why AI, why not AI, the good, the bad, the naughty, the fun, the acceleration, the unavoidability of it all. Just getting on the Highway 101 from San Francisco to Palo Alto is enough to get the conversation started...
In the end, I always come back to a 3-legged stool...
1. The Problem.
We need to ensure that we are fixing a real problem, one that will bring something to the table. What problem? that is really up to each organization to find out. Sometimes it will be staring in our faces; in other situations we may want to dig down... but AI should be engaged to solve something, not just to say "we're doing AI".
Not only it should be a real problem, but we should understand what the solution will bring, and how we will measure success. Examples...
Mandate: "fix recruitment with AI" is the wrong approach; do we know what is wrong? fixing recruitment could be a good hunch, but to engage AI in the right way we need to do homework, understand what we want to fix (costs? quality of hires? time to hire? all three?). In some cases, AI can be a support and an accelerator. In other cases, only a costly add-on, and there are other options.
I suggest some Design Thinking to dig deep...
The crowning glory of finding the real problem to solve is that it facilitates adoption. If you give folks a spoon, they will use it to eat the soup. If you give them a fork, they'll continue sipping from the bowl.
2. Data.
Of course, analytics, reporting, AI are nothing without data; and in the case of AI, enormous amounts of data. Once we have identified the real problem to solve (and only then), we can review what data we need to have as a foundation, and determine if we have it, if we can procure it (either sourcing internally, or buying externally), if we need to enrich it...
Data must be pertinent, and relevant to our organization, industry and geography. Data on hiring in - say, Nairobi - is not very useful to determine pay scales in Rome, Italy.
... and finally, the big one:
3. Governance!
This is not trivial at all, and cannot be just sourced somewhere. One of the biggest challenges is that there isn't "recipe", and one must examine carefully the multiple parts to determine how to approach AI governance in your organization (and yes, I am underlining "your" because it has to fit the culture and vision).
Then, after examination, it is about assembling all the different parts, understand how each fits in your environment and in the fast evolving legal environment around us.
A place to start is linked; a well rounded report, addressing multiple facets and offering good suggestions.
