The hype is real! Well, maybe… Over the last few weeks I don’t think there’s a leader in the Global 2000 and beyond that hasn’t been asked or thought,
“What can ChatGPT or any of the latest hype of AI technologies do for our business?”
or almost just as likely,
“Just how risky are these technologies?” .
Both are important questions, and rarely answered with an applied AI or practical outcome lens. More importantly the tools and processes we’ve been using to introduce technology has already taught us the broad brush strokes of managing risk, creating committees, and connecting teams, etc. Lovingly, these create great slide decks, but seldom result in moving the business forward for new technologies; instead they bias to observation and risk management. To take advantage of AI there needs to be a change in how enterprises assess and get AI into production.
The real role of AI leadership
Today the majority of ownership for AI lands with IT, InfoSec, HR, and other usually horizontal teams (source: Baker McKenzie). It’s understandable this happens, AI is nerdy, creates risks, and is complex to understand. It doesn’t help that technology companies explaining and selling AI struggle to start with a business outcome, and then for their sins try translate any question as quickly as possible into you purchasing their platform.
Said no CEO ever, “I’m so glad I have the perfect AI platform! My strategy is done”
Don’t get me wrong, I’m guilty of this sin in previous lives, but AI leadership needs to step back from technology, and instead begin to think in new ways about how to drive practical value. This can only come by identifying transformational use cases that start with a business statement.
The real role of AI leadership is to understand the most critical business problems in ultimate detail, and translate those into a portfolio of AI initiatives for the organization, while managing C-Suite buy in and alignment, and reporting back the performance of each investment.
So right about now you’re probably asking, why does this need to be a new C-suite role? Surely a CFO, CIO, CTO, or another role should be able to effectively achieve this. Maybe, but AI has been on executive radars for the last 10 years, yet according to a Deloitte’s state of AI report in 2022, 50% of projects lack executive buy in (source: Deloitte), and we’ve all heard some version of a stat about just how few AI projects make it to production. I would argue this is due to organizations using traditional methods to assess and progress AI adoption. There’s probably a sprinkle of Conway’s law between these organizational dynamics as well, meaning company wide transformation is doomed to fail from the start. To kick start AI transformation, the AI charter needs to report directly to the CEO and be responsible for bringing all senior leaders together in a common plan.
So what next? You’ve decided to establish the Chief AI Officer role, what should they do? How do we ensure that we’re not just creating a bunch of expensive slide decks or experiments that don’t reach production? As a leader how do you measure success of this role?
There are four common dimensions to track to understand the success and impact for adding a CAIO to your business:
Company-wide AI transformational & initiative roadmap — This the hardest step to aligning the organization. You’re probably thinking, of course it is, but hear me out. We’ve been doing roadmaps wrong this whole time. Usually the process of creating an AI roadmap starts with a fancy conference room, some blueberry muffins, a bunch of ideas, putting them into a deck, and then presenting them six weeks later to a room of people that mostly agree with you -FAIL. Instead invest in the roadmap to include transformation programs from education and AI projects themselves, but also let this become the framework to align all teams with, establish principles for prioritizing and scoring the impact of initiatives, and then educate, communicate, and in many cases add backlogs of new projects. Make the roadmap review and ideation a regular event.
Transformation roadmaps only work if there’s clear principles to make changes and approve new projects, and ultimately decline them too. Without this approach, give it time, and you’ll have 5 company-wide AI transformational roadmaps, one from each team that didn’t get what they wanted in the original version.
Impact Register & transformation status — So projects are underway, there’s progress reports, but accountability is key. Starting an AI project is easy, finishing one is hard.
A good CAIO upfront calls out the desired impact, and then continuously focuses the organization to deliver that outcome, in production. A great CAIO, then tracks the real impact of that project and consistently updates the business on the performance of that project over time.
Just as important to measure success is ensuring that the overall organizational evolves with and alongside the initiatives. Don’t under estimate the impact of antibodies resisting the change to the status quo. As you can see from the AI generated image above, some of those antibodies might be right. As a leader showing both the projects that are ahead or behind schedule, alongside the progress in organizational change management will be informative in both directions. Remember there’s a reason it worked the way it used to, take the time to understand the resistance at the shop-floor, chances are you’ll be able to build even better AI with that information.
Competitive landscape — This one gets forgotten, and underestimated often, just ask our friends in Mountain View.
As a member of the C-suite, this escalates the expectation of the CAIO to not only drive transformation and impact, but also effectively understand, and communicate how competitors may leverage AI to disrupt your business, or add new competitive pressures.
A successful CAIO is able to provide an exposure and control model for competitors potential initiatives, but also create counter-measures or contingency plans should they happen. Remember it’s not necessarily preempting every scenario, but being able to recognize and respond to potential threats in future.
AI Program Management — This is the most common focus on AI transformation today. We all recognize how important managing bias and fairness, audit-ability, etc. are to program management. But wait there’s a little more. Successful AI program management when led by a CAIO needs to extend program management, up and down the organization. This means a regular program update in board meetings, and regular team updates with the broader organization. Company-wide transparency on initiatives, and an education program for all levels of the business.
Program management needs to expand into demystifying how AI programs can support the business and their teams. Successfully doing this means getting there faster, but most importantly, I’m going to bet $1 that during those training sessions across the company, you’re going to discover your next big AI idea.
See it’s not that hard after all. Joking aside, I get it, this is a lot. But nothing that is worth it is easy. And you can wait for the easy button, but then what’s to stop someone else from figuring it out before you?