As I think most of us are doing, I recently had Copilot give a read to communication that I drafted before I sent it. I treat AI as non annoying spell checking this way, and it sometimes catches things I'm grateful for.
However, Copilot in particular is prone to unraveling a thing I've been working on in this organization for ten years, and I have no intention of allowing that to happen. Put simply, I'd ask you to remember this:
Do not cushion reality by default. In this environment, clarity is safer than comfort.
Here's what I mean by that. Copilot tends to optimize written word for buffering reality to make things easier to hear. Ask it as much; it'll confirm. And depending on what it is that you're saying, that can be a good thing. The line I set for myself is to be gentle about people, and uncompromising about reality.
Copilot will often suggest wording that softens the nature of a problem to make it easier to hear. I would ask you to read for this and to consider the following. When we soften the reality of a problem, we often slow the urgency and response actions. The consequence of that is two fold:
- Important things don't get addressed when they should, and more importantly
- Anyone who's willing to overstate reality will get inside our decision loop, and force us to make bad decisions.
Think about that for a moment. Have you ever seen a situation where responses to problems don't make sense? If you think back, does the pattern I describe here resonate with what had happened?
Let's avoid that. We should practice kindness, and that often means saying things that might make someone unused to candor bristle. I would suggest the solution to that is building a culture in which such statements are normal and welcome, rather than something to avoid.
One last thing; I'm not advocating that we forget our 3VQ training and become the Persecutor. Such behavior is more caustic towards an open relationship with each other than any buffering AI may bring. Remember, AI is a tool to be used with judicious eyes.