Pascal and I had an unsettling experience recently while running a scenarios workshop with one of our clients. We’d been asked to help a group of leaders explore a set of scenarios envisioning how the development of digital technologies over the next decade might impact demand and fulfillment in their industry.
In the early steps of this workshop, each group is engaged in worldbuilding – working out from a few high-level defining facts about their respective scenario to create an increasingly detailed and imaginative but logically consistent, plausible version of the future. One of the groups, tasked with envisioning a world in 2033 where artificial intelligence had proven to have a “transformational” impact, began modestly: They suggested that this future would see AI/automation start to replace human labor in a range of jobs and functions.
A part of me is always relieved and grateful when the human imagination doesn’t immediately jump to Skynet or Agent Smith when asked to envision a future defined by transformative AI tools and systems. But this version of 2033 was looking almost shockingly mundane, strangely familiar, and probably deeply defined by an unexamined assumption that tomorrow would still look mostly like today. Real change would still be something on the horizon. There would still be time to prepare for the “transformative” part of the future to arrive.
Scenario thinking is designed specifically to help us break free of that kind of assumption, free of the “tyranny of the present” and the hold that past experience has over the futural imagination. And there’s a strong case to be made for the claim that this type of thinking (exploring “what we might do if…”) is particularly essential now and particularly urgent as regards the development of AI systems. You don’t need to be a card-carrying singularitarian in the mold of Ray Kurzweil to recognize that the field isn’t just moving fast – the rate of change is clearly accelerating. The next generation of multimodal generative AI tools (which can process, analyze, and create across media types) is already leaving ChatGPT behind – barely six months after the chatbot’s public launch.
The thing to realize here: Precisely because all of these new tools and technologies are digital, they are far easier to iterate, share, combine, fork, embed, and remix – each act contributing to the larger process of combinatorial innovation – than were the defining technologies of previous eras. All of which means that the transformational future is moving fast and could arrive on a timeline that will feel deeply unfamiliar.
Do I know what the trajectory of AI will look like over the next decade? Of course not. That’s a tremendous time frame for these technologies to evolve and converge. But I know this: We need to give ourselves, our leaders, and our organizations permission to imagine TRULY transformational impacts on the economy, society, and culture. Seeing those possibilities with clarity will allow us to ask richer questions and frame better, more intentional conversations about stakes and strategy, how we might need to lead, and what we might need to learn. (via Jeffrey)