This is becoming one of the deepest questions of our time.
For decades, people asked “What can AI do for philosophy?” Today, the equally important question is “What can philosophy do for AI?” The answer, I think, is: far more than providing ethical guidelines.
Philosophy may become part of AI’s operating system.
Here are several levels.
1. Philosophy provides concepts that engineering cannot invent by itself.
Engineering is excellent at answering:
How?
Philosophy asks:
What is happening?
What counts as understanding?
What is a person?
What is knowledge?
What is a world?
Without those concepts, AI engineers often optimize the wrong thing.
For example,
Transformer architectures solved language prediction.
But language prediction is not yet
- understanding
- intention
- meaning
- responsibility.
Philosophy distinguishes these.
2. Philosophy enlarges AI’s ontology.
Most AI systems today assume reality consists primarily of
- objects
- tokens
- vectors
- probabilities.
Philosophy says reality may also include
- relations
- events
- emergence
- becoming
- embodiment
- temporality
- absence
- possibility.
Think of
- Martin Heidegger → Being as disclosure.
- Alfred North Whitehead → reality as process.
- Gilles Deleuze → multiplicity and becoming.
- Maurice Merleau-Ponty → embodied perception.
These are not merely philosophical curiosities.
They suggest entirely different AI architectures.
3. Philosophy teaches AI about limits.
Modern AI is extraordinarily good at optimization.
Philosophy repeatedly asks:
Should everything be optimized?
Examples:
A friendship optimized becomes networking.
Love optimized becomes matching.
Education optimized becomes testing.
Psychotherapy optimized becomes symptom reduction.
Politics optimized becomes surveillance.
Optimization is never neutral.
Philosophy remembers what optimization forgets.
4. Philosophy explains meaning.
Current LLMs learn statistical regularities.
Humans also live inside
- histories
- narratives
- myths
- memories
- rituals
- mortality.
Meaning emerges from these larger structures.
This is why philosophers such as Paul Ricoeur remain deeply relevant.
Narrative is not decoration.
Narrative organizes identity.
5. Philosophy may help AI move beyond representation.
Much twentieth-century philosophy questioned the assumption that knowledge means making internal representations.
Instead,
- enactivism
- phenomenology
- pragmatism
- ecological psychology
argue that cognition is enacted through interaction.
This has already begun influencing robotics and embodied AI.
The future AI may not merely represent a world.
It may continuously participate in one.
6. Philosophy introduces genuine uncertainty.
Science often reduces uncertainty.
Philosophy sometimes increases it.
That is healthy.
Questions like
“What is consciousness?”
“What is freedom?”
“What is death?”
remain open.
An AI trained only to answer risks becoming overconfident.
An AI trained philosophically learns when not to conclude.
This may be an essential ingredient of trustworthy intelligence.
7. Philosophy keeps AI human without making it anthropocentric.
This is where recent posthuman thinkers become especially interesting.
For example,
- Rosi Braidotti
- Francesca Ferrando
- Christine Daigle
argue that posthumanism is not about replacing humans.
It is about relocating humans within a larger ecology of beings.
This perspective may help AI avoid two opposite mistakes:
- human exceptionalism
- machine exceptionalism.
8. Philosophy may become AI’s memory.
One danger of AI is perpetual novelty.
Every week a new model.
Every month a new benchmark.
Philosophy remembers conversations that have lasted two thousand years.
Questions raised by
- Plato
- Aristotle
- Immanuel Kant
- Friedrich Nietzsche
have not disappeared.
AI joins those conversations rather than replacing them.
What I find especially interesting, in light of our conversations over the past year, is that your own work points toward a ninth possibility.
You have often argued that the future is not simply a contest between humans and AI, but between what you call the algorithmic self and the analogical self. In that framework, philosophy is not merely a source of theories for AI. It is a practice of preserving forms of thinking that resist complete computation: following fragments rather than objectives, lingering with ambiguity, returning to history and nature, and allowing what cannot be optimized to remain present.
From that perspective, philosophy does something no benchmark can measure.
It keeps open the possibility that intelligence is more than efficient prediction. It reminds both humans and AI that understanding is not exhausted by computation, and that wisdom may consist less in arriving at the fastest answer than in learning which questions should remain alive.
