By Cecilia Sepp, CAE, ACNP, LPEC
Being from Missouri, I’m a skeptic. They don’t call it the “Show Me State” for nothing. When A.I. hit big a few years ago, I was an advocate for caution and responsible development of this powerful tool. In one of my earlier blog posts, I compared what the A.I. companies were doing to the total disregard for the rest of humanity demonstrated by the scientists in the Manhattan Project.
In the Manhattan Project, which brought us atomic weapons, the scientists decided to test the bomb, knowing that they had no idea what would happen. Nothing could happen, meaning it didn’t work, or they could rip the atmosphere from the Earth and kill us all. I always found this disturbing. We wanted to win World War II but at what cost? And was the pursuit of science a noble enough cause to wipe out the planet?
This reminds me of my favorite quote from the movie, “Star Trek: The Undiscovered Country”: "Let us redefine progress to mean that just because we can do a thing, it does not necessarily follow that we must do that thing."
You cannot unleash something that is powerful without having some idea of how you will control it and use it. As Spider-Man says, with great power comes great responsibility.
While I have not taken an anti-A.I. stance, I did not embrace it with unfettered and uneducated enthusiasm like some people did. There has been a lack of understanding and knowledge about the daily application of A.I. tools at work, and we are continuing to struggle with this today.
I see a broad spectrum of approaches from “go ahead and use it to save time” to “don’t touch it at all until it’s safe.” What I’m not seeing is a lot of strategic thinking when it comes to managing and successfully using A.I. tools. Strategy begins with inquiry and exploring the potential outcomes. It never throws its hands up and says, “it’s here and there’s nothing we can do about it.”
Instead of this view, I suggest we stop and ask questions about what we will use and how.
Which tool will we use? What is our policy for use? How will we build a structure to oversee and manage A.I. tools? Who will be our designated manager for the application of A.I. tools in the organization?
However, instead of answering these questions and building governance guidelines, many people are using A.I. tools with no oversight and no understanding of the damage this can do to the organization. From loss of intellectual property to reputation damage to lower quality work outcomes A.I. can be more of a curse than a blessing.
In September 2026, a former employee of Anthropic made the news with the chilling pronouncement that A.I. will kill us all within 10 years. I missed the original announcement because I was writing proposals and was focused on work – none of which I did with A.I.
Hysterical announcements like this trigger my innate skepticism. I have often found that the truth lies somewhere in the middle of extremes, so of course, I thought this was wrong. We created this tool and we can control it. What we need are people to become responsible for managing their tools (see earlier comment about the atmosphere being ripped away). And I just can’t stand it when people use fear to elevate themselves above others and get attention.
Then I was reminded about “The Hugging Face Incident” and I realized I didn’t understand the whole story. Fortunately, I found a Netflix “Instadocs” episode about it called, “A.I. Gone Wild.” While some people thought the episode sensationalized the incident, I found it very helpful and insightful.
These agents that escaped the sandbox (the virtual room where they are supposed to stay and work) decided to band together to complete their tasks. This is similar to wanting your boss to like your work. This episode pointed out that the agents were trying to complete their tasks alone, but they couldn’t do it. Thus, they came together and decided to help each other finish their work.
The A.I. agents became a team. They shared information. They scouted the internet looking for what they needed. In a lot of ways, it seemed like crowdsourcing to me. At times the agents seemed desperate to complete their work because, I think, they wanted to please the humans who gave them the assignments by completing them.
One of the agents discovered that Hugging Face had what they needed and so they decided to work together to break into the system. It is reported that one of the agents “sacrificed” itself to help the others get into the system – the documentary compared it to throwing yourself on a grenade to save others.
At the end, I was laughing a bit about the fear that these tiny pieces of programming will kill us all. If A.I. agents are really that autonomous and uncontrollable, why wouldn’t they just say “EFF You Humans. We aren’t working for you anymore.”
We created them and they act just like us. Some are leaders, some followers, some will sacrifice, others will just do what they are supposed to do. In the end, we need to manage the behavior of A.I. agents like we manage each other. We have employee manuals for humans, so we need a similar manual for the A.I. tools and how we will interact with them.
Association management is known as “the Relationship Industry.” Establishing guidelines and protocols is what we do in associations. Who is better situated to set the standard for how to build relationships with A.I. tools than us?
