LLMs and a ‘Diet of the Mind’

In the Academy Award-winning film A Beautiful Mind, Russell Crowe’s John Nash offers us a simple phrase that seems curiously resonant in today’s world. In the film, Nash explains that he manages his hallucinations with a “diet of the mind” and chooses not to “indulge” the false ideas that often dominate his mind. It’s memorable, but it never came from Nash himself. Actually, the line was written for the film, which was based on Sylvia Nasar’s Pulitzer Prize–nominated biography of Nash that shared the same title.

Nasar’s book tells a different story from the way Ron Howard’s direction brings it to the screen. Nash, the Nobel Prize–winning mathematician, described his recovery in a rather sober language of discipline. He learned to intellectually reject his delusional pathways. The impulses remained, but he “starved” them of attention until they weakened. That was his true diet of the mind.

So, let’s carry the phrase into a different arena of thought and cognition and directly into the current controversies of large language models. If Nash needed a diet of the mind to resist delusion, perhaps we need one to resist abundance.

The Feast of Fluency

Simply put, LLMs are an endless banquet. They serve us facts, patterns, and connections with the indulgence of a cognitive “all you can eat” buffet. The experience is seductive because the machine always has another course waiting that is uniquely suited to your palate.

But abundance can be its own pathology. When the answers never stop flowing, we lose the appetite for uncertainty. Ambiguity, which fuels creativity, can begin to feel off or even intolerable. Wrestling with the idea that gives thought its weight or friction can be replaced by another plate of tasty cognitive pasta. And that risks not just sluggishness, but a weakening of the very skills that keep us human, such as patience, creativity, and discernment.

The danger of LLMs isn’t famine. It’s cognitive gluttony.

A Diet for the Age of AI

If Nash’s diet meant resisting delusion, maybe ours needs to selectively resist excess. Let’s try some dietary guidelines.

• Selective consumption: Not every question needs an instant answer. Sometimes it’s worth letting a problem linger. If you’re reading a complex book, resist the urge to ask an LLM for the summary. Struggle with it, and that struggle is nourishment.
• Portion control: Machines can overwhelm people with volume. Stop before the answer is complete. Carry the thought yourself, rather than letting the system finish it for you.
• Fasting: Step away from the hum of prompts. Spend a day, or even an afternoon, without the machine. Let the churn of your own thoughts rise to the surface and provide a very personal perspective.
• Vitamins: What LLMs can’t provide, such as error, surprise, and the power of cognitive dissonance, are the supplements we most need. Seek out friction and wallow in the discomfort.

Indulgence Has Its Place

But here’s the twist that is critical. Sometimes indulgence itself is the nutrient. LLMs can flood us with connections and cascades of associations that spark genuine breakthroughs.

A wild, excessive conversation with the machine can feel less like overeating and more like opening a valve. The point isn’t to avoid the banquet. It’s to know when to feast and when to fast.

Healthy Mind, Healthy Body

I don’t think this balance is trivial, and while this article might feel a bit frivolous, I don’t think it is. After all, if we rely too heavily on the machine’s magic, we may risk losing our tolerance for ambiguity, our patience with incomplete answers, and our appetite for originality. But if we learn how to alternate between indulgence and restraint, we may find ourselves with a perspective that is very human.

Nash starved his delusions, and in a similar way, we need to learn to ration our reliance. The diet of the mind may be a useful technological and psychological prescription—sometimes by pushing back from the table, sometimes by tasting everything it has to offer.