0

Let me explain what I mean. From the standpoint of biology, life is often defined by having a metabolism, the ability to reproduce, evolve, etc. (There is no single definition so far as I can tell.)

Likewise, "intelligence" has an array of definitions, but the etymology strongly implies "selecting between alternatives". All definitions of intelligence are arguably rooted in "fitness in an environment", a utilitarian lens arising out of theory of evolution, where an environment is any "action space" (here defined as a system that can produce outcomes), with degree of intelligence merely a measure of fitness, typically versus other decision-making mechanisms.

In the same way a discrete value has no meaning except in relation to other values, intelligence has no meaning except in relation to a "problem" or context in which a decision or action can be taken and an outcome produced. (Many game theorists would argue that any set of choices always reduce to a binary--a given choice vs. any number of alternatives, so even a single choice becomes a binary--that single action vs. no action.)

Further, taking an action or making a decision requires the mechanism to be "animate" i.e. active, as opposed to inanimate i.e. inactive. (The other core definition of intelligence is merely "information", so the primary distinction may properly be active vs. inactive information. As an example, the code for an AI stored in a text file vs. that same code processing and making decisions.)

  • Is it therefore valid to consider any mechanism that takes any action a form of life?

In other words, any such mechanism, if observed, is demonstrating fitness for environment, just as life by any definition is subject to fitness in an environment. There's also a theory that everything is information because the phenomenal world comes to all organisms & mechanisms through a filter of perception. Thus matter is ultimately information in a specific form.

  • Sorry if this is convoluted or not fully formed--it's a notion I've been considering and welcome all perspective. (I can clarify or expound at request.) – DukeZhou Mar 25 at 20:56
  • "Taking" an action and producing a reaction is called causality. You use a very subjective meaning of the term "decision" (decisions require reason, but you cannot suggest that a crab is rational). A rock can "take" an action as receiving a lot of energy and "decide" (causal consequence) to break or not. But a rock is not alive. Anyway, intelligence is an attribute of an abstract object, reason, and it is not really related to life, which is an attribute of physical objects. Your question can be improved by removing your opinions and being more precise. – RodolfoAP Mar 25 at 21:29
  • 1
    As with most words, etymology has little to do with the modern meaning, and is not an argument for the latter being analogous or related to something else. Insects are extremely well fit to their environment, many survived almost unchanged for much longer than "intelligent" mammals existed. Degeneration of intelligence can even produce more fit species, as in the case of parasitic organisms. So no, intelligence is not a measure of fitness, if that is what "analogous to life" means. The idea of "evolutionary progress" in intelligence or complexity is generally rejected in modern biology. – Conifold Mar 25 at 23:21
  • 1
    If there is a lot of confusion about "intelligence" then what is "the modern conception"? It seems to me that you are simply redefining it as "fitness to the environment" to create the "analogy", which is not what the usual use is, despite the confusion. Even then it does not work if "fitness" is taken in the usual biological sense of successful gene propagation. – Conifold Mar 27 at 2:41
  • 1
    I don't understand how this is a question. Algae is alive but no one would say it has intelligence. – curiousdannii Mar 27 at 7:44
2

If one uses the definition of analogy as a criterion, then obviously yes the two are analgous. They are unlike, and thus not identical, but can be compared because they have features in common. Most obvious is that both are understood as teleological but in different ways. It bears mentioning that you might want to peruse the article in the SEP: Teleological Notions in Biology. There's a big difference between the lungs having the "goal" of exchanging gasses through a biphasic interface of gas and liquid, and a student having the "goal" of graduating high school. It would be very easy to fall into a fallacy of equivocation, so beware.

Let's grant the traditional views of biological life and intelligence because it can get to be a sticky-wicket if we equivocate on terms, esp. intelligence. A thermostat can be called an intelligent device, but using these terms to describe artificial systems of computation requires caveats we needn't deal with here. Also, be careful of the fallacy of black and white thinking. Things needn't be intelligent or not. That's childish thinking. Rather, to what degree is something alive or thinking is much more sophisticated and saves a lot of useless metaphysical pedantry.

Now, the short answer is that life is viewed as teleological generally without awareness, but awareness in a system grants it the descriptor intelligence. Metacognition would be teleological too, but with the added quality of self-awareness.

Think of it this way. A cell has a host of homeostatic requirements regarding pH and salinity and metabolic processes. A cell is clearly alive, and we can say it senses it's environment, but to consider this intelligence is the narrow, physically-oriented sense of deterministic causality. For instance, a simple one-celled organism doesn't cogitate like a bonobo about bananas and sex; rather, it's like a simple machine that senses and detects and we can trace the causality through the system.

The teleology of an organism like a dog is much more sophisticated because a dog can learn in the classical sense that psychological learning theory demonstrates. It has a sophisticated body composed of tissues and systems such as the nervous and endocrine systems which lead to non-deterministic behavior, which can be understood statistically, but not so much at a physical level, but rather at a biological level. How we group our concepts of 'things' is what ontology is all about, and so to contemplate metabolic cycles and observable canine behavior are built on two different primitive classes of ontological primitives. A biochemical compound comports with atomic theory, and a dog's body comports with evolutionary theory, and a dog's behavior comports with psychological theory, roughly.

So, both life and intelligence are systems that can be viewed through the lens of analogy, but like all analogy, it depends on how much you want to abstract. Let's list some examples of analogous features:

Both life and intelligence:

  • obey naturalism; neither are supernatural
  • conform to the theories of physics, chemistry, and biology
  • can exhibit complexities not found in mere particles, atoms, and molecules; have emergent properties
  • both are negentropic and obey thermodynamics
  • both can take inputs, give outputs, change themselves internally, affect their environments, and self-replicate
  • both can be represented and simulated with symbolic abstractions
  • both are concepts which are abstractions of their subsystems

I could go on, but these characteristics are sufficient to qualify the claim that life and intelligence are analogs of a sort.

Is it therefore valid to consider any mechanism that takes any action a form of life?

The answer to the second question is no. A digital thermostat can take action by turning on and off a furnace according to a very sophisticated set of computations, but it is not alive because it lacks some of the other characteristics we ascribe to life including self-replication and growth.

|improve this answer|||||
  • Well argued. I consider a digital thermostat, and even a basic switch, alive, so long as they have a power supply. (Albeit, a switch would the most basic life-form;) It's interesting that we even talk about such circuits as being "live" when they have current. Also interesting that human brains use electricity. Is a single biological neuron alive? – DukeZhou Mar 27 at 0:40
2

Yes to analogy. No to conclusion.

Probably the closest to an agreed definition of life is systems manifesting autopoeisis, the ability of a system to maintain and reproduce itself. The unit of replication is a complex topic

Intelligence can broadly be linked to fittedness and resilience, and it is almost a lucky accident that at a certain point our brains which had developed for social complexity had other payoffs enough to justify the expense of using 20% of our calories, as now. Intelligence can, and often is unconscious, a classic example is stotting, a behaviour by springbok exhibiting high levels of fitness, read by lions as guidance not to pursue. Broadly intelligence can be characterised on a continuum with other aspects that contribute to fittedness and resilience. In this sense AI algorithms are like biological algorithms like viruses - and evolutionary algorithms explicitly mimic trial and error of units in a test environment, and replication with variation, so are another step.

But you ask about intentionality, and that relates to consciousness. Specifically, to have intentions is to model oneself in the landscape of choices, and to have some kind of self concept to do that with. Douglas Hofstadter's 'strange loop' model is capable of distinguishing systems without intentions from those that do, by looking at how self-concepts create feedback loops and tangle knowledge hierarchies. This also relates to subjectivity, and the hard problem of consciousness. Why aren't we just composites of algorithms, why do we unify what happens to us in the experience of subjectivity? Hofstadter's approach says it is related to having a level of complexity that means a system can hold itself as an element in it's cognitive field.

So there is a hierarchy. Complex systems (chemistry, code). Complex systems with replication and variation (viruses, life, genetic computer algorithms). Complex systems capable of modelling themselves in creating options to pick their actions from (animals that pass the mirror test, artificial general intelligences).

Evolutionary algorithms have replicators within them, but the whole system doing that is not a replicator. The computer can't make new copies of itself. Something like Alpha Zero pushes that further, it is closer to qualities of life (and behaves similarly to how brains build reinforced 'pathways'). Next will be von Neumann machines, a machine capable of building a copy of itself. People already try to do something similar by making 3D printers on 3D printers - there will come a time that can be automated, and be done using available raw materials. With variation of each generation, that would be more like life still. To have intentions is a higher-level behaviour where that replication process allows some kind of feedback involving a self-model.

|improve this answer|||||
  • Fully agree with the analogy of current algorithms to microbial life--modern software is analogous to biological colonies (man-o-war, etc.) I agree with Hofstadter's recursive theory of self, and the chaotic system it produces, but we're still making assumptions about free will (we only know we have the appearance of it.) So I can't base anything on the assumption of non-deterministic intentionality. What I mainly arguing is that mechanism (computation) is itself (active) intelligence, regardless of fitness, and "I compute, therefore I am". – DukeZhou Mar 27 at 0:35
  • Thanks for this excellent answer, btw. You clarify that in my conception, autopoesis is not a requirement, only mechanism. – DukeZhou Mar 27 at 0:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.