There is an idea that intentionality may be a requirement of true intelligence, here defined as human intelligence.

But all I know for certain is that we have the appearance of free will. Under the assumption that the universe is purely deterministic, what do we mean by intention?

(This seems an important question given that intention is not just a philosophical matter in relation to definitions of AI, but involves ethics in the sense of application of AI, "offloading responsibility to agents that cannot be meaningfully punished" as an example. Also touches on goals, implied by intention, whether awareness is a requirement, and what constitutes awareness. I'm interested in all angles, but was inspired by the question "does true art require intention, and, if so, is that the sole domain of humans?")

  • 3
    $\begingroup$ Under the assumption the universe is deterministic, I would say you can't argue humans are intelligent either. Without free will, how are we any different than a complex algorithm that has many responses to different inputs? $\endgroup$ – Recessive Feb 28 '20 at 2:13
  • 1
    $\begingroup$ On the contrary, the difference between a deterministic universe and a random one wouldn't be noticeable on the macro scale, so that brings up the interesting argument of why should something that doesn't have an affect on the macro scale all of a sudden mean some things (on the macro scale) are no longer possible (in this case true intelligence)? None the less, I believe AI is capable of exhibiting the same level of intent that humans do, whether that be "fake" (no free will) intent or not. $\endgroup$ – Recessive Feb 28 '20 at 2:17
  • 2
    $\begingroup$ See Free Will and Incompatibilist (Nondeterministic) Theories of Free Will (two Stanford Encyclopedia of Philosophy articles). $\endgroup$ – nbro Feb 28 '20 at 19:33
  • $\begingroup$ Free will is religious concept, not scientific. $\endgroup$ – mirror2image Mar 3 '20 at 6:16
  • $\begingroup$ can you source the claim that " intentionality may be a requirement of true intelligence"? this seems to be an artifact of simon-newell era goal-directed intelligence definition, but surely aesthetic appreciation counts as a mode of intelligence? what about emotional intelligence? $\endgroup$ – k.c. sayz 'k.c sayz' Jul 30 '20 at 3:48

My thoughts.

The short answer is: you can't.

The long answer is that since we're searching for a new definition of a term when removing a necessary (in my opinion) precondition for it to exist, the question becomes "can you make up a new definition for what you intuitively and empirically understand as intention, while removing free will from the picture?". I'm going to give it a shot.

First of all, there's a lot to be said about whether or not the idea or intuition of intention even exists in our collective discourse based on the latent assumption that free will is a thing. As in, before any rigorous definition, even the intuitions encoded in discourse, philosophy, art, and other languages as "intention" could very well be as invalid as the assumption of free will itself.

That being said, I'm a fan of Deleuze's model for people and other entities as machines of input, output and internal state (not his wording, but I paraphrase and in consequence, interpret and alter for the purposes of my point). It's not perfect, but I run to it a lot to answer these question as I find it very refreshing, often lacking in bias and having good explanatory power compared to the usual romance-foo that dominates these conversations. If that's the case you could pretty much define intention not as a self-started force but as a product of a much blurrier mechanism, namely the non-deterministic characteristic of this rhizomatic soup. Whether or not an input or output will exist, what kind will it be, what internal state will it find or cause in the machine and the long term dependencies between these interactions seem to me like a convincing enough candidate for the cause of any intuition (or illusion if you like a more cynic vocabulary) of "intention". It's pretty much an emergent symbol we use, assuming the form of a force for the setup and function of new connections, that will in their complication or pure non-determinism spawn even more intention in the network.

tl;dr: Intent could be the most basic expression of the RNG of the universe.

  • $\begingroup$ Thanks for this excellent answer and welcome to SE:AI! I'm strongly persuaded that, although we cannot yet model the human brain, it is some kind of computer, in that it takes input and typically produces output. It seems to me that intention can merely be a expression of pursuit of some goal, thus intention does not require free will or even non-deterministic process. Bell's Theorem strongly indicates randomness in nature, so I also like your point about it's potential relationship to intent, precisely because a coin flip is the opposite, unless its part of a search function. $\endgroup$ – DukeZhou May 10 at 0:12
  • $\begingroup$ @DukeZhou I like the way you approach it, I also think that if probabilistic versions of typical macro-level physics can in some way hold truth then it seems that intention is a sort of inefficient or probabilistic search for a predetermined goal, delineated by our specific universe (as opposed to any counterfactual). $\endgroup$ – Nikos Tsakas May 10 at 8:07
  • $\begingroup$ Of course we are again constrained by the language we use, since in a way the ability for a system to have a goal in the first place is what I took to be a product of free will, which is why I had that whole intro. If goals can occur as contributions to some universal one, then of course no free will as we understand it is required but my personal intuition of what a goal is kind of breaks down at that point and it seems like we are talking more about a ball rolling down a rocky hill and search for paths down instead of choosing to fly vertically simply because it isn't concerned with gravity. $\endgroup$ – Nikos Tsakas May 10 at 8:07

In order to answer the question requires that the definition of intelligence (human or otherwise) be grounded into the physical model. There are two forms of intelligence in the physical model, one subconscious (general) and the other conscious (specific).

The subconscious is responsible for managing the brain’s process and performing all the physical reactions. It exist in a deterministic universe where the decisions are known and reactions automated. In the brain, the shape of the deterministic universe is time. For something to be deterministic, it must have guaranteed results within a specific time cycle. This requirement precludes the use of reasoning and long-term memory whose completion time cannot be fixed. The subconscious runs on general intelligence. To understand the format of general intelligence, think in terms of a jigsaw puzzle. General intelligence only needs the border of the puzzle. With just the basic object structure, the subconscious can move and process data throughout the brain without regards to its content. It is functionally agnostic and can complete its automation cycle without reasoning.

Now, consciousness is responsible for mapping the non-deterministic environment to the deterministic universe. To do this, it must exist outside of that universe so its own non-deterministic nature doesn’t interfere with the subconscious process. Consciousness runs on specific intelligence. Its job is to decipher the contents of the puzzle. In humans, specific intelligence takes on the form of visual symbols. These symbols are used by the conscious state for reasoning and long-term memory access. How decisions are reached has no predefined form or technique. It is a non-deterministic process based on the accumulation of individual experience and previous selection. This process allows the biological life-form to adapt decision-making to its own environment. In the most basic sense, the subconscious allows us to live in a deterministic universe and consciousness allows us to adapt to a non-deterministic environment.

The two forms of intelligence are functionally incompatible. The non-deterministic environment has no common translation to a deterministic universe. If it did, it wouldn’t be non-deterministic. To overcome the translation problem, the lower occipital lobe is a functional “Black Box” that allows the two different formats to be written into the same memory construct. By simple association, the two systems coexist in a non-invasive relationship where the hippocampus (short-term) memory serves as an interchange point.

With a grounded physical model, I can now attempt to answer some of the questions.

  1. We have “free will”, but it can only exist in the non-deterministic environment of the conscious state. There is no such thing as “free will” in a deterministic universe. The results of that universe are already known and decided.
  2. A deterministic universe cannot change. Its own nature precludes adaption from within. To make alterations requires a separate state that can rise above the process in order to change the process. Biological consciousness serves this role and requires “free will” to make decisions outside of the scope of what is already known.
  3. There are no ethics in a deterministic universe. Abstracting right and wrong is a reasoning function which is not allowed. AI constructs like agents which do pattern matching to develop predictions/reactions are not directly attached to any conscious state. They are deterministic functions whose capability is predefined. Simply put, agents cannot exceed the sum of their programming because they cannot adapt to that which is not known or anticipated.
  4. Liability only exists where “free will” exists and does not exist in the deterministic universe. Otherwise, you are attempting to blame your hand for a decision originated in your frontal lobe.
  5. “Free will” becomes a prisoner of the deterministic universe. As experience and decision knowledge grow, more and more decisions are shifted to the subconscious. This creates a dependency that restricts the range of “free will” and the decisions it is called upon to resolve.
  6. Creativity does not exist in the deterministic universe. It is an essential skill used by the conscious state to employ related memory to formulate new decisions. Creativity is used by animals to develop survival skills to adapt to environmental changes. The production of art work in an esoteric sense is purely human, but as a basic function is not.
  7. Intent does not exist in the deterministic universe. Within the non-deterministic universe of consciousness, intent is buried in an ocean of previous experience that may or may not reflect current “free will” goals. Especially, if these goals threaten the survival of the biological process.
  • 1
    $\begingroup$ Can you please cite any research work that supports your claims about the subconscious, conscious, general intelligence, specific intelligence, etc.? I've never heard of these theories, you'd better just cite the theories behind your claims. $\endgroup$ – nbro Mar 1 '20 at 18:26
  • $\begingroup$ Sorry, I can’t cite any research other than my own. The information I supplied in this answer was provided from an Artificial General Intelligence prototype that I have been developing. The design of the prototype is based on a neurological algorithm derived from the physical model. Actually, I would have been stunned if you had heard anything about this approach. My thanks for confirming that belief. I am about to release a paper describing the full algorithm possibly later this week. If you are interested, I can circle back around and provide a link. $\endgroup$ – David Edgar Mar 2 '20 at 1:48
  • $\begingroup$ The following two papers explain the brain’s physical model and basic algorithm. I have found by using this model, I can begin to answer some of the hard questions such as “free will”. It makes things less mystical, more systematic. medium.com/@davidedgar_5686/… medium.com/@davidedgar_5686/… $\endgroup$ – David Edgar Mar 6 '20 at 15:47

The term "intentionality" has two quite different senses. One is a very technical concept in philosophy of AI and means (roughly) aboutness in the sense that beliefs, desires, fears, etc., are about things (snakes, tax, chocolate). That's the sense central to searle's Chinese room argument. The other sense is the common idea of intending to do (or not do) something. A really key issue for AI according to Searle is how can a computer have intentionality in the first sense.


I like both answers but am going to propose something simpler:

  • Intention can be understood as an expression of pursuit of a goal, and thus does not require free will.

Deterministic algorithms can have and pursue goals, even without being consciously "aware" of the goal. Thus, a simple NIM solving algorithm can be said to have the intention of winning at NIM, even there the goal is not a part of the algorithm, but embedded undefined in the simple algorithm itself.

This becomes even more true with neural networds, where, unlike the NIMATRON, goals are typically defined.


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.