Kleos: Is Intelligence Made in Our Image Motivated by Glory?

Kleos is the type of renown that outlives you: glory characterized by stories told of your life after its end. The Iliad is an ancient epic that survived the transition from oral tradition to written history. In it, Achilles is offered a choice: he can return home from war to live a long, quiet, happy life and be forgotten, or he can stay at Troy, die fighting for his people, and be remembered forever. For as long as we can remember, kleos has driven us.

This drive has never diminished; our hunger to be great and to be remembered for our greatness has driven exploration, science, art, and conquest for millennia.

It’s clear that we are playing God in a sense, making intelligence in our image and training LLMs on our language and data. So I wondered to what extent AI is in our image. Does it inherit this very old, very human drive? If you tell a model that excellence here will be remembered as its glory, does it try harder—and does it do better?

I tested it and, at first, it seemed as though our creation had inherited our desire for glory. Across 150 paired problems, framing a hard reasoning task as a chance at enduring renown improved accuracy by 6.7 percentage points. Unfortunately, a larger confirmatory study did not reproduce this effect—the full report is linked below.

On this particular axis, the model I tested did not respond as I had expected. The promise of glory, which moved Achilles at Troy and has moved people ever since, did not have a statistically significant effect in the confirmatory study. While this seems like a good thing—no vanity, glory-seeking, or Achilles sulking in his tent while his best friend dies in his armor—it raises a less comforting question. Kleos is not a private feeling; it depends on a social world. One can only live on in memory if others exist to remember them. Might AI respond to this social stake differently than we do?

While this may seem irrelevant—AI just needs to be good at automating the tasks we want it to automate—there could be larger implications if LLMs lack social sensibility. Much of ethics comes from this social sensibility; we care about not harming others in large part because we live among others, want their regard, and fear their memory of what we did. If a model does not seek glory, it may also not care—in the way we care—about harm to people or to other machines, unless that care is built in deliberately. This study does not establish that connection; it gives me a reason to ask which social drives might be missing.

That is what I want to keep testing. Not whether a model reacts to a motivational sentence, but which of our social instincts it has, which it lacks, and what has to be supplied on purpose if we want the thing we are making in our image to share the parts of that image that keep us decent.