Notes for 3/16/2026

  

3/16/2026

[Philosophy Club every Monday, 4-5 pm, in the Buchtel College of Arts and Sciences room 436 ("The Cave")]

[Bioethics Club: Mondays from 5:30pm-6:30pm in Leigh Hall 408]

 

What’s the worst thing you’ve ever eaten?

Leibniz’s mill thought experiment:

 

…It must be confessed that perception and that which depends upon it are inexplicable on mechanical grounds, that is to say, by means of figures and motions. And supposing there were a machine, so constructed as to think, feel, and have perception, it might be conceived as increased in size, while keeping the same proportions, so that one might go into it as into a mill. That being so, we should, on examining its interior, find only parts which work one upon another, and never anything by which to explain a perception. Thus it is in a simple substance, and not in a compound or in a machine, that perception must be sought for. (Leibniz, Monadology §17)

 

What WOULD “explain a perception”? What does a perception *look like* to anyone who isn’t having it?

 

LLMs can now (mostly) pass the Turing Test

Does this mean they think?

Do they understand?

 

 

 

There’s nothing inside an LLM that looks like understanding.

 

But neither is there anything inside a brain that looks like understanding.

 

 

Identity theorists and functionalists think Leibniz’s mill represents a kind of mistake.

(But what kind of mistake, exactly?)

 

 

What is it to understand fear, for example?

 

Me: “What makes a movie scary?”

Claude 4.6: “A movie becomes scary by exploiting the gap between what we know and what we fear — using darkness, silence, and slow-building tension to make audiences feel genuinely unsafe. The most effective horror corrupts something familiar (a home, a child, a reflection), withholds information to let imagination run wild, and strips characters of control so the audience feels helpless alongside them. Sound design and pacing do much of the heavy lifting, priming the nervous system with quiet dread before delivering a shock — but the films that truly linger are the ones that tap into something personally or culturally resonant, making the threat feel not just cinematic, but real.”

 

Does Claude understand what “scary” means?

How could it when Claude has never experienced fear?

 

 

When we know how a LLM works, it seems easier to dismiss it as really intelligent because we’re essentially inside Leibniz’s mill; nothing looks like understanding.

 

 

One line of disagreement concerns different ways of defining “understanding.”

 

 

Horizontal vs. vertical understanding

(syntactic vs semantic meaning)

 

Basic LLMs have only horizontal understanding: words connect to other words (or perhaps letters to letters, or perhaps only numbers to numbers).

 

The important point here is that AI systems like this have no connection to the world, which is what a “vertical” form of understanding would involve.

 

More recently, however, AI systems have been designed that are capable of image processing and recognition. AI systems control robots that can move and apply a variety of sensors (“senses”) that could constitute vertical understanding.

 

 

But no AI systems currently have the capacity for a vertical understanding of emotions or other subjective states that humans experience.

 

 

Remember the discussion of mental images from last week?

 

If mental images exist, then I/O functionalism can’t be right. (And note that they can’t be “seen” inside Leibniz’s mill.)

 

This connects with a resurgence of interest in the subjective that we’ll talk about next time.