Getting Ahead of the AI Adoption Curve

It’s estimated that 10% of IT budgets will be spent on AI in 2024. Yet very few IT departments have a strategy for AI adoption. Fortunately, we can apply some of the lessons from the SaaS boom to get ahead of the AI adoption curve.

New AI Tech, Same Old Problems

Roughly a decade ago, software as a service (SaaS) hit IT departments like a runaway freight train. The SaaS sprawl boom saw enterprises eagerly adopting literally hundreds of applications to meet highly specialized business needs. This proliferation of applications, while beneficial, also led to increased security vulnerabilities, compliance overhead, and difficulties in managing and integrating multiple systems. These challenges assaulted IT on three fronts:

1. People skills – Before the SaaS era, many IT professionals were accustomed to managing on-premises systems. For example: IT had personnel that could manage Windows and Linux servers, but professionally managing complex SaaS operations in Salesforce, ServiceNow, and Office365 was a completely different ballgame.
2. Technology – Governing SaaS at scale required new technology, which wasn’t introduced until long after that SaaS boom. Solutions like SaaS management platforms and cloud-native software asset management systems were late to the party, and were primarily purchased to react to ungoverned SaaS sprawl. 
3. Process – Business-led SaaS initiatives catapulted hundreds of SaaS apps over the traditional IT castle walls of due diligence. The problem here was really two-fold: IT’s manual software review processes couldn’t scale fast enough to onboard the rapid acquisition of new applications, and secondarily, the business units themselves weren’t talking to each other, which led to wasteful overlaps and redundancies in software functionality. 

Enter the current era of AI, and it’s easy to see that history is repeating itself. Many different AI solutions with overlapping functionality and highly variable degrees of governance are hitting organizations like a scattershot cannon just as we saw with SaaS a decade ago. The good news is: we’ve learned a thing or two about rapid technology adoption, and we can leverage those lessons to decrease the pain in this rapid tsunami of AI adoption. 

What Governed AI Adoption Looks Like

Governed AI is a journey, not a destination. The important thing is to start immediately with something, then iterate and improve. Again, the goal here is to get ahead of the reactive chaos by setting up the foundational building blocks that would put IT in the driver’s seat rather than the back seat. These building blocks include:

1. Process – Design your enterprise AI adoption framework. This is your overarching master plan that will holistically establish where IT plays in AI across the entire company. It will also establish guidelines to provide autonomy to business-led AI selection so that IT isn’t on the critical path of every minute decision. 
2. Technology – While the business is preoccupied with the AI project of the week, IT needs to be instrumenting observability, security, and governance platforms to manage the mass acquisition of new AI tooling. 
3. People skills – Establish role maps for your IT staff whereby you assert the types of roles needed first. For example: AI prompt DLP analyst or non-human identity management engineer. These new roles require new skills, so flesh out the roles with requisite training modules via your internal learning management system and set training goals for staff to accomplish in a time-boxed fashion. Many IT workers of today complain of lack of opportunities for growth or promotion, so here is a perfect avenue to train and promote from within!
4. Leverage outside expertise – While AI vendors have professional implementation partners, they’re typically focused on implementing that single solution. To understand how to manage the macro-picture of AI at scale, you’ll likely need a consultant to assist with getting your AI adoption program off the ground. 

Why IT Leaders Should Care

Seasoned IT leaders should be familiar with the “IT orphanage” axiom whereby poorly managed technology eventually finds its way to IT. In a nutshell, tech is purchased by a business unit, then managed somewhat sufficiently at a small scale. When adoption increases, the business unit soon realizes it’s performing a wide range of IT-like functions such as appeasing the security team with technical controls, performing DR and BCP activities, and so on. The business unit then realizes these are IT tasks, and promptly punts the solution over the fence to IT. This of course is unappealing to IT, as the IT department is already swamped with work and likely would not have implemented the in-flight solution as it stands today. 

Governing AI (or any technology for that matter) across the enterprise is what’s best for the company as a whole, yet there’s an IT-centric motive here as well. If you’re not proactively governing the entire AI footprint across the enterprise, you’ll instead operate within a reactive state; managing AI on an app-by-app or project-by-project basis. This transactional, non-strategic state is the antithesis of modern IT and there’s really no way to escape the bad image this will reflect on your IT department.