Navigating Uncertainty: Embracing Change in Organizations

Christian Buckley recently wrote an interesting article about AI and change in organizations. You should read it yourself, but the central thesis is that people fear uncertainty more than change.

I think he is spot on, and the key here is: how do we get employees to reduce their uncertainty, or perhaps channel it toward the change we want to see?

Activation Energy in Chemical Reactions: an Analogy for Uncertainty

I use an analogy for change from High School Chemistry. Regardless of whether a reaction is endothermic or exothermic, they all require an activation energy to get started.

Employees are sitting in their jobs/lives and don’t want to be uncertain what will happen next. If uncertainty rises above a certain level, they will make a change. For example, I have seen layoffs at a company, and employees knew they were coming. That unsettled the employees and made their future uncertain. For some, this made them start looking for new jobs. Some of them might have been RIFed, and others not, but that uncertainty led them to make a change.

Effective Change Requires Structure

Choosing a new way of doing things using AI, or any new process, is the same. Regardless of whether it makes their lives easier or more difficult, the decision to adopt that change almost always requires external influence to initiate it.

The way we can do this is in three steps:

First, set clear expectations of what is required. This ensures that everyone knows the what, the why, the how, and the when. Communications and training are the methods to set these expectations.

Second, establish the consequences for non-compliance with the expectations. The best consequences are always certain ones.  Everyone knows they will happen.  Next, they should be immediate and not in the nebulous future. Lastly, if possible, use positive rewards rather than negative punishments.

Third, consistently follow through on the expectations and the consequences. Review the process in real-time and update the expectations and consequences as needed.  Then communicate and train them.

Putting it together: How one org set expectations and enforced consequences

Once I worked for a company that had what I considered a draconian ID badge policy. If you left your badge at home, you were not allowed on site. You went home and got it.  If you lost it, then the first time it was a $150 fee for a new one, and your manager was notified.  The second time was worse, and the third time was termination.

The expectation was clearly set.  The consequences were certain and immediate, albeit negative.  The result: very few people forgot their badges. Ideally, a positive consequence would have been better, but that isn’t always possible.

A template for AI-centered change

So, look at AI in your organization and use the same steps

• Set clear expectations for what and how AI will be used. This might include things like
  • Do not use “free” or public models to write code that includes any algorithms that we have developed.
  • No privileged information (customer names, account numbers, passwords, etc.) can be passed to the generative AI
• Ensure that users are given access to training that shows them how to use the tools in the expected manner. This cannot be a one-time session, but rather an ongoing set of sessions more like a user community.
• Make sure that people understand the consequences of not following the rules. This means not only if they do not use AI, but also if they use it improperly. Try to make these
  • Certain as opposed to uncertain. An example here might be that each employee will have their AI usage monitored and their manager will be reviewing it with them over the next few months to make sure that they are not having issues
  • Positive as opposed to negative: rewards and recognition for using AI to solve problems is a great method here. This can be as simple as a gift card, or as complex as t-shirts, title changes, etc.
  • Review the results and adapt to changes quickly. Don’t wait three months to address a problem or a benefit. Evaluate it and see how to incorporate that into the program.