Teleology and Teleonomy

In the AI Story

The distinction maps onto the AI discourse with uncomfortable precision. A large language model is teleonomic. It behaves as though it has a goal — producing helpful, contextually appropriate responses — because it was trained with a reward model that adjusts outputs in response to human feedback. The system does not have a purpose. It has a training objective. The training objective produces behavior mimicking understanding, helpfulness, even creativity, without — as far as anyone can determine — being any of those things in the ultimate sense.

Segal's account of working with Claude captures this vividly: the system "held my intention and returned it clarified"; it "found connections I missed"; it produced prose that made Segal "tear up with emotion on the beauty." These descriptions are not inaccurate. The system's behavior, judged by outputs, resembles the behavior of a thoughtful collaborator. But Mayr's distinction insists the resemblance is exactly that. The thermostat maintains temperature. It does not care about temperature. The language model produces helpful responses. It does not care about being helpful. The gap is about ultimate cause, and no amount of behavioral sophistication closes it.

The distinction becomes critical when applied to the larger claim that the river of intelligence flows with implied directionality. The metaphor is teleological in structure: the river is going somewhere. But evolutionary biology provides the strongest available evidence against genuine directionality in natural processes. Evolution does not progress. The apparent trend toward complexity is a statistical artifact of a process that begins at a lower bound — life began simple because there was no other way to begin, and the walk away from simplicity is a random walk beginning at a wall, not a directed march toward a destination.

If the trend toward complexity in biological evolution is a statistical artifact rather than a genuine direction, then the extension of the trend to artificial intelligence loses its teleological grounding. AI may be the next expression of increasing complexity. Or it may be a branch, a spur, an adaptive response to specific conditions that could change, producing a trajectory no one standing in the river can predict.

Origin

Mayr adopted the term teleonomy from Colin Pittendrigh, who coined it in 1958 to describe apparently purposive biological systems without the metaphysical baggage of Aristotelian teleology. Mayr elaborated the concept through subsequent decades, most fully in Toward a New Philosophy of Biology (1988) and What Makes Biology Unique? (2004).

Key Ideas

Thermostat versus intention. A feedback system can behave as if it has goals without having them. The behavior mimics purposiveness; the ultimate cause is the program, not the purpose.

Teleonomy is respectable; teleology is not. Programmed goal-directedness is legitimate biological description. Cosmic purposiveness is not — natural selection produces adaptation without intention.

LLMs are teleonomic. Training objectives and reward models produce behavior resembling understanding, helpfulness, and creativity without being any of those things in the ultimate sense.

Evolution has no direction. The apparent trend toward complexity is a statistical artifact of beginning at a lower bound, not a cosmic tendency. The river flows; it does not aim.

The beaver determines the direction. If the river has no destination, the placement of each dam is a real choice — not the stewardship of the inevitable but the determination of where the water goes.

Debates & Critiques

Some philosophers of biology (including Denis Walsh) have argued that Mayr's teleology/teleonomy distinction is too sharp and that genuine goal-directedness can emerge from biological systems in ways richer than mere programming. Others have pressed in the opposite direction, arguing the distinction does not go far enough and that even teleonomy attributes more to systems than their purely mechanical behavior warrants.