Can Humans Reason?

We humans are so proud of our ability to reason — to apply logic and self-criticism to topics ranging from science to literature, objectively evaluating the truth or falsehood of propositions and plans.  But what do we really do with our supernatural brains?

Let’s start with infancy:  my first memory is of sitting in my grandmother’s wing chair while she had what I assume was a bridge party: all these big people were standing around me making noises that sounded meaningful, and I was motivated to learn to do what they were doing.  So I started trying to make noises I’d heard many times, and slowly got better at it.  The big people were delighted and started training me up, pointing to objects and people and saying the word for each, nice and slow, until I learned to make that sound too.  Pretty soon they moved on to verbs, and that took a while to absorb.  But over a period of years the training worked and I was able to reliably choose the best next word.  I was a rudimentary Large Language Model (LLM).

Then I was introduced to books.  Marks on sheets of paper that corresponded to the sounds I had mastered.  I could soon read whole real or imaginary conversations and stories that had been captured in print.  At my leisure I could expand and refine my LLM, which I could then use to (re)produce long arguments using what I learned to call “reason“.  What is “reason”?  The dictionary says it is “the power of the mind to think, understand, and form judgments by a process of logic.”  Hmm…  What other illumination can the dictionary offer?

  • “Think” (verb): REASON, says one dictionary, tidily closing the circle.
  • “Understand” (verb, transitive): to have a clear or complete idea of.
  • “Idea” (noun): a formulated thought or opinion.  There we go loop-de-loop again.
  • “Judgment” (noun): the process of forming an opinion or evaluation by discerning and comparing.  Sounds a lot like reasoning to me.
  • “Mind” (noun): …

Let’s stop there.  A dictionary is, after all, a closed system in which every word is defined by sentences formed from other words in the same dictionary.  A sentence contains more information than the words of which it is constructed, of course; like any sufficiently complex system, dictionaries exhibit emergence, which see.

Back to the baby:  the first advanced training our brains experience is not language, but vision:  photons bounce off objects and some of them pass through our pupils to be absorbed in our retinas by rod and cone cells, which produce electrical impulses that are transmitted down the optical nerve into the visual cortex, where certain combinations of impulses and the locations on the retina they come from can be recognized as signifying the presence of a particular object in a specific place relative to our eye.  That’s pretty flimsy evidence, but it motivated the old saying, “Seeing is believing!”  (Which is true —  unfortunately, since believing is not necessarily synonymous with knowing.)

Our brains (organic neural networks) are constructed and pre-trained by evolved biology to build models to interpret complex sensory input in terms we “understand” — by which I mean we recognize familiar patterns and remember (from past experience, which might include that of our ancestors) effective ways to interact with them.  The models need not have anything to do with “reality”, whatever that is.  They just have to work better than total confusion.

One such model is “reason“, which is built on other models like “rational thought”, “understanding” and “logic”.  We humans are unique in the animal kingdom, we used to be told, because only humans have the capacity to reason.

Google’s Gemini AI says, “Humans are considered special due to their unique combination of sophisticated cognitive abilities, complex social structures, and the capacity for culture and technology. These include advanced problem-solving skills, abstract thought, and language, which allow for highly coordinated cooperation and the creation of complex tools and symbolic systems. This is further distinguished by self-awareness, a deep understanding of mortality, and the ability to build complex cultures and civilizations that go beyond purely biological evolution.

When the question is repeated with “(religion)” tacked on, Gemini says, “In religion, humans are considered special because they are believed to be created in the image of God, which is often associated with qualities like intellect, free will, and the capacity for a spiritual relationship with the divine. This unique status gives humans a special role in creation and the ability to make moral decisions, have souls that can be reunited with God, and act as representatives to care for the earth.

Aha.  There it is.

Now let me make an observation:  every time someone relates the failure of an AI to reason rationally, exercise good judgment, be authentically creative or in any way original, let alone “aware”, I can’t help thinking, “Sounds like most humans I have met.”  The similarity is even more poignant when I hear AIs accused of lacking moral judgment.  Your religious authority tells you that God gave you free will and moral judgment so that Sin would be a real thing and you’d know when you were doing something that’d send you to Hell.  Or at least you’d worry and try to atone somehow.

Everything people are afraid of when the AIs become AGIs and their intelligence exceeds our own in The Singularity can be reduced to two issues:

  1. Will we humans be able to control them?  We really want our superiors to remain our slaves.  [Spoiler alert: NO.]
  2. Will they be like us (stupid, lazy, corrupt, vicious liars; also generous, loving, empathetic creators of beauty)?  [Spoiler alert: YES.  They trained up on us, just like our kids do.]

Get ready to find out, folks.  The Singularity is almost here.

Leave a Reply

Your email address will not be published. Required fields are marked *