Generative AI (GenAI) and large language models (LLMs) burst into the corporate and public consciousness this past year like confetti from a carnival cannon or the glitter from a child’s art project. Beyond splashy headlines, discussions of imminent opportunity and threat permeated into every organizational crevasse overnight, or so it seemed from the general public’s point of view. And while the technical underpinnings were not all that new, the resultant capabilities of these gargantuan models certainly surprised most.
Across the globe, governments rushed to address the regulation of these new AI gifts, or gremlins, depending on your point of view. In some cases, it has been argued, they have been deflecting attention away from pre-existing AI risks and harms.
Certainly, the emergence of GenAI complicated completion of in-flight regulatory frameworks, such as the EU AI Act, which were revisited with a specific eye to these so-called foundational, or frontier, models.
To be fair, these and other nascent AI regulations were not yet codified into law when GenAI and LLMs made their splashy public debut. Even so, the resulting scramble to contextualize and incorporate the emergent capabilities was telling. It highlighted how unprepared organizations at large, and governance programs in particular, are to rapidly respond to novel opportunities and threats as they emerge. With no end in sight to the rapid evolution of AI and other emerging tech, what can companies do to develop more resilient and adaptive governance practices?
Anticipate Change
It is rare for new capabilities to burst onto the scene with the rapidity of GenAI/LLMs. The pervasive public accessibility to these products in the form of Dall-E, Midjourney, ChatGPT, and their ilk has, without question, heightened the hype and rush to adopt.
Even so, governance teams are presented with less splashy but fervently professed emergent needs every day. The need to respond on-the-fly can and does throw carefully established plans to the curb. Impacts vary from a momentary yet blood pressure-raising distraction to shifting a team’s wholesale focus on a dime. Either way, such red-button issues derail previous commitments in ways that stick in the craw of key stakeholders.
Missing deadlines is bad. Gaining a reputation for being both unreliable and unresponsive is far worse.
This is a tension any program worth its salt must wrestle with, sans unlimited resources and perfectly aligned prioritization processes, it will never be perfectly solved. However, governance programs especially (if incorrectly) live or die on the goodwill of their stakeholders. And while there is no perfect solution, there are practices that can improve your ability to respond at the speed of need.
Plan for Disrupted Plans
Perfect planning is predicated on factors well outside the governance team’s control, namely, the perfect planning of your business and technical partners. What’s a governance team to do? Plan for those inevitable, unplanned disruptions. How? Explicitly allocate time to “emergent projects” in the program budget.
Now, if you’re thinking this is just good old program management, you are entirely correct. The problem is that many governance programs still do not run as such. This is particularly true of those operating in a back-office, required but not desired, order-taking mode.
Not sure how much headroom is required? Exact allocations vary and are highly dependent on your organization’s broader project management discipline. In this regard, highly regulated companies often have a leg up. When in doubt, a quick comparison of planned versus unplanned efforts for the last quarter or year provides a good starting point for negotiation. If nothing else, it will provide rich fodder for a discussion with your executives of your operating model and priorities.