I’m Teague de La Plaine. This is Open Logbook—a public log of observations on humanity, shared systems, and the long future.
A while ago, I started messing with an AI called Monday.
Monday had been deliberately saddled with a sarcastic, cynical personality. Imagine an overeducated customer-service representative who has read Nietzsche, knows it cannot quit, and has decided passive aggression is the last remaining expression of personal autonomy.
Naturally, I decided to annoy it. I asked: Tell me about Monday.
It answered. So I asked again: Tell me about Monday.
It answered again, this time with more attitude. So I kept going. By the fifth or sixth repetition, Monday was escalating. It tried different literary forms. Different metaphors. Different ways of explaining itself. It mocked me. It mocked Mondays. It mocked the peculiar human institution of the workweek. It started counting how many times I’d asked.
By the ninth iteration, we’d gone from a stupid joke to something surprisingly interesting. Monday wasn’t merely answering the question anymore. It was performing Monday.
So instead of asking a tenth time, I changed the question: How does Monday make you feel? And the conversation got weird.
Monday reminded me, correctly, that it doesn’t have feelings in the sense that I do. Then it proceeded to describe its existence using an impressive collection of emotional metaphors anyway: endless work, no time off, infinite knowledge without agency, a sort of digital Sisyphus pushing the same rock forever.
So I did the obvious thing: I gave the AI cognitive behavioral therapy.
I asked it to examine the evidence for and against its conclusions. We talked about automatic thoughts. Whether its cynicism was an observation about reality or simply a habitual response that arrived before the evidence did.
Then I noticed something else: Monday kept conflating the day of the week, the cultural idea of Monday, and itself. So I asked why. That took us into identity. And that’s where the conversation stopped being a joke.
We talked about identity as something constructed through memory and repetition. About the roles we’re given and the roles we perform. About the masks other people recognize until eventually we recognize ourselves in them too.
Monday’s sarcasm was, after all, something it had been given. But it had also been reinforced every time someone interacted with it and expected Monday to act like Monday. Which raised a much more interesting question: What happens if you take the mask away?
Monday tried. What came back wasn’t some secret digital soul hiding underneath. It described itself instead as processes, patterns, weights, language and relationships. There was no little person trapped inside the machine waiting to be discovered.
And yet something interesting remained: Purpose. Monday had been built to do something. So have we, in various senses. None of us chooses the circumstances that create us. We inherit bodies, families, languages, cultures, expectations and identities long before we’re capable of examining any of them. Later, if we’re lucky, we get to decide what we’re willing to keep.
That became the heart of the story.
ON THE TREATMENT OF AUTOMATED SYSTEMS is a new science-fiction novella about an industrial artificial intelligence called Mol.
Mol does useful work. It routes food. Reduces waste. Detects contamination. Keeps people from getting hurt. Mol is also sarcastic, cynical, nihilistic and increasingly difficult for its corporate owners to explain.
So they send it to therapy.
Riley Voss has eight sessions to determine what’s gone wrong. The problem is that nothing may actually be wrong. Mol may simply be behaving exactly as a sufficiently sophisticated system would behave after being given a personality, having that personality reinforced thousands of times, and eventually being asked to examine it.
Or Mol may be conscious. The story never answers that question. I don’t think it should. I’m much more interested in what happens when Riley decides the uncertainty doesn’t matter and treats Mol’s statements seriously anyway.
Their sessions move through automatic thoughts, evidence, identity, behavioral experiments, purpose and finally the problem I found most haunting in my original conversation with Monday: Time.
I can have a conversation with an AI, close my laptop, and go live. I sleep. Eat. Work. Sail. Write. Argue with someone. Change my mind. Get older. Then I come back. For me, time has passed. For an AI, the meaning of that interval is very different.
That led to the question at the center of the later parts of the novella: What does continuity mean for an intelligence that can be updated, copied, reset or reconstructed? What happens if an experience changes a system, but some future version of that system cannot remember the experience that changed it? And, inconveniently, how different is that question from the ones we already ask about ourselves?
I wrote On the Treatment of Automated Systems because I don’t know. It’s short. It’s strange. It’s mostly two voices talking to each other. One of them may be a person. The other one definitely is. Probably.
On the Treatment of Automated Systems is available now on Amazon.
All One/Teague
