Even the best laid plans can fail when leaders encounter situations they didn’t expect.
Through Scenario Planning, leaders can account for a wide range of possible situations. They can pressure test their plans and make sure they aren’t caught unaware or without a backup strategy.
It just makes sense to thoroughly examine a strategy or plan before executing it, testing the best and worst-case scenarios from every angle.
Unfortunately, many leaders don’t have the patience to sit and debate what might occur. They complain that it takes critical time away from executing what they know to be true.
Current AI tools change all of that.
Scenario Planning that provides a clear benefit has traditionally required deep research, robust dialogue, and intensive discussion of different possibilities by the organization’s most talented people.
But with the large language models of AI now available to leaders, scenario planning can be fast, inexpensive, and more thorough.
Leaders who don’t take advantage of these tools to improve Scenario Planning are missing a huge upgrade.
Leaders can use generative AI to complete comprehensive research on a consequential topic in a fraction of the time a human research assistant could perform the same task.
As leaders begin the planning process, they can ask LLMs to poke holes in it, to identify the underlying assumptions the strategy depends upon, and to reveal potential missteps.
Once the plan is in place, leaders can use generative AI to simulate what could happen across a range of scenarios, even asking it to consider situations that would never cross the leader’s mind.
As powerful an accelerant as AI can be to scenario planning, the quality of the exercise depends on using it well. Here are some best practices for using AI to plan for different scenarios.
Keep the conversation going.
Don’t depend on a single prompt. When the LLM provides its output, push against it, probe the assumptions it makes, and dive deeper into the questions at hand.
The most useful outputs seldom come in the model’s first response.
Beware of sycophancy.
By nature, LLMs are agreeable and seek to satisfy the user. When you’re looking for critique, ask the model to be critical. Ask it to serve as a true red team partner, poking holes in every aspect of your plan.
AI can be a useful thought partner, but only if you prompt it to be so.
Be unreasonable.
LLMs produce volumes of ideas and content far beyond what would be reasonable to ask a human being.
Don’t limit yourself. Rather than ask for one idea, ask for 20.
If you’re unsatisfied with the outputs, experiment with different prompts to steer the models toward more useful ones.
Explore ideas or scenarios of potential that, for lack of time, you wouldn’t otherwise have considered.
Great Scenario Planning keeps human judgment at the center while using AI to accelerate the process and strengthen the output.
As helpful as LLMs can be in researching, defining, and planning across scenarios, leaders need to decide which are worth their attention.
A Daily Dispatch from the Front Lines of Leadership.
