Michael, you make a very interesting point when you say, "we attach quantities to what we perceive. I cannot perceive two things without at the same time making a judgment as to whether they are ‘same’ or ‘different’ (which may be construed as elementary quantities, like 0 and 1). It may have required some training, but I cannot perceive a train of syllables, like con-cu-pi-sci-ble, without assigning places for them in an order and counting their quantity. But Claude does not work in this way. Just as it does not construct formal models, it does not assign quantities in this way." Question: You point out a fascinating deficit of Claude's underlying LLM structure. My question is regarding your reference to Plato. I thought that St Thomas Aquinas following Aristotle would not combine our perception of two objects as difference and quantity at the same time but instead would assign a value difference as a result of intellectual activity rather than from raw sense perception of an image? Also, given the deficit in Claude's underlying structure, is this deficit the agentic gap since it is the soul that does this activity in the human person? Or, is/are my question(s) flawed?
Peter, I believe that intelligence informs human perceptual activity as a rule. What you refer to as “raw sense perception” is an abstraction from actual experience. Yet Plato was making a stronger, metaphysical claim, that it is simply not possible for any form to be participated in (even in ‘raw sense perception,’ if a form can be found there) except by that form’s also participating in Sameness and Difference. As to Claude, I’m still trying to figure out what’s “off” about it and keeping in mind the question whether these deficiencies can ever be remediated. I suspect they can be patched in most cases, but that the patches would not be efficient.
Michael, you make a very interesting point when you say, "we attach quantities to what we perceive. I cannot perceive two things without at the same time making a judgment as to whether they are ‘same’ or ‘different’ (which may be construed as elementary quantities, like 0 and 1). It may have required some training, but I cannot perceive a train of syllables, like con-cu-pi-sci-ble, without assigning places for them in an order and counting their quantity. But Claude does not work in this way. Just as it does not construct formal models, it does not assign quantities in this way." Question: You point out a fascinating deficit of Claude's underlying LLM structure. My question is regarding your reference to Plato. I thought that St Thomas Aquinas following Aristotle would not combine our perception of two objects as difference and quantity at the same time but instead would assign a value difference as a result of intellectual activity rather than from raw sense perception of an image? Also, given the deficit in Claude's underlying structure, is this deficit the agentic gap since it is the soul that does this activity in the human person? Or, is/are my question(s) flawed?
Peter, I believe that intelligence informs human perceptual activity as a rule. What you refer to as “raw sense perception” is an abstraction from actual experience. Yet Plato was making a stronger, metaphysical claim, that it is simply not possible for any form to be participated in (even in ‘raw sense perception,’ if a form can be found there) except by that form’s also participating in Sameness and Difference. As to Claude, I’m still trying to figure out what’s “off” about it and keeping in mind the question whether these deficiencies can ever be remediated. I suspect they can be patched in most cases, but that the patches would not be efficient.