There is something I find fascinating about carbon monoxide. The gas is molecularly simple (one carbon and one oxygen), colorless, and odorless. Yet once it enters the blood, it takes on a “life” of its own by binding to hemoglobin with more than 200 times the affinity of oxygen. And interestingly, carbon monoxide poisoning is generally associated with a cherry-red or pink appearance, which is an almost perverse contradiction where this toxicity can carry a color we associate with health and vitality.
It sounds like something from Gabriel García Márquez, where magical prose combines with bitter almonds, poison, and a touch of unrequited love. There is something beautiful and deceptive about the physiology, which may be why this idea has resonated with me since my days as a firefighter and paramedic. Oxygen is what the body needs, yet carbon monoxide is simply better at the binding site.
My contention is that there’s a simple yet powerful idea here. High affinity does not necessarily mean that something is good for a person or system. Lately, I have been wondering if this has something to teach us about artificial intelligence.
When AI Competes With Thought
Let’s start with the idea of cognitive work. Do I do it—think about it or write it? Or do I hand it off to an LLM and get the answer in seconds? The result may often be similar, but the workflow isn’t.
In this context, AI seems to have an extraordinary affinity for “doing the work” for us. AI is fast and even fun. And in a biological sense, this begins to look a little like a type of competitive inhibition because AI can occupy that cognitive space where “effortful thought” once was our “cognitive oxygen.” Simply put, AI competes with our brains.
It’s important to recognize that this doesn’t mean that AI always diminishes cognition. I have often had the opposite experience and I expect that you have, too. My idea of iterative intelligence came from those exchanges where an answer from AI becomes the start (not end) of a power dynamic.
Something Else May Be Happening
Carbon monoxide gives us another interesting perspective. Its effect is broader than just binding tightly to hemoglobin. Once CO is bound, it changes the structure of the hemoglobin molecule, making bound oxygen harder to release to tissues. The interaction of CO with hemoglobin affects the behavior of the larger system.
Could it be possible that our interactions with AI begin to influence how we think? If answers are always there at our fingertips, maybe uncertainty itself finds a new and higher level of intolerance? And if well-crafted solutions arrive at the touch of a button, maybe our human cognitive path begins to feel less critical or even important.
And yes, I’m using biology here as an analogy. I’m not suggesting some hidden receptor deep in our neuroanatomy. Still, this possibility fascinates me because the effect on us could be new—on some fundamental level. It’s no longer the “quid pro quo” relationship of handing a task to a machine and getting the answer back. Noncompetitive inhibition describes how the broader and larger function of a system is altered by the battle for a receptor site. I wonder whether there’s a cognitive correlate to that idea worth exploring?
AI can compete with our act of thinking. But repeated interactions may do something more subtle. It may begin to change our propensity to engage in “effortful thought” in the first place. A type of cognitive inhibition.
What AI Does With Us
Maybe this is part of what makes AI “feel” different from earlier tech. A calculator adds a group of numbers, and search finds information for me to review. Both are more transactional than iterative. But an LLM can enter our thinking while the thought itself is still taking shape.
And that can happen before I have decided what I was trying to say in the first place.
Of course, that can be enormously useful. There are moments when the AI’s coherence sharpens and expands our own thinking. But there are also moments when I wonder if AI’s answer arrives just too soon. The thought looks finished, yet my cognitive task remains somewhat jilted.
Much of the discussion around AI has focused on what the technology does FOR us. I’m becoming more interested in what happens when AI begins doing things WITH us, participating in thought while it is still emerging and influencing what comes next.
The Part We May Not See
Cognition adapts to the environment (think neuroplasticity). And AI is, without a doubt, becoming part of that environment. I don’t know if repeated use produces anything that deserves the name “noncompetitive cognitive inhibition,” but I think the question is worth asking.
What if AI does more than compete with thought and begins to change our propensity to engage it?
Here’s my hyperbolic take-away. CO gives you that rosie look of oxygenation. And AI gives you that look of cognitive authority. But all may not be as it seems. The biology reminds us that the thing that binds most readily is not always the thing the system needs. And that’s something worth thinking about.