Project Q*, OpenAI, the Chinese Room, and AGI

If anyone desires it, I can tie this in to grade-school mathematical reasoning—explaining how mental matchsticks and the like can keep arithmetical, algebraic, and geometrical LLMs over their targets—but for the moment, my bet has been placed.

User: I wonder if in some ways, the forms or ideas in Plato can be understood as being more like principles. I think there is a tendency to depict them as something very like 'shapes' whereas there are ideas, even in the very early dialogues, which are plainly not like shapes (like 'the form of largeness' in the Parmenides).

ChatGPT: Your observation about the Platonic Forms being more akin to principles than to physical "shapes" is quite insightful. Indeed, there is a common misconception that Plato's Forms or Ideas are somehow tangible or have a specific physical shape, largely because of the term "Form." However, in Plato's philosophy, Forms are indeed much closer to abstract principles or essences.

The Platonic Forms are not physical entities; they are non-material, transcendent, and immutable. They represent the perfect, unchanging essence of things. When Plato refers to the Form of a tree, for instance, he is not speaking of a superlative, physical tree existing in another realm, but rather the very essence of "tree-ness," the principle or concept that defines what it means to be a tree.

In dialogues like the "Parmenides," Plato explores more abstract Forms like "largeness" or "beauty." These discussions make it clear that Forms are not physical shapes. "Largeness" doesn't have a size, and "beauty" doesn't have a specific appearance. Instead, these Forms are the fundamental realities that underlie and give meaning to the various manifestations we encounter in the sensory world. The Form of Largeness is the essence of what it means to be large, the principle by which we can recognize something as large.

In this sense, understanding the Platonic Forms as principles or essences is more accurate. They are the standards by which we measure and understand the myriad of things we experience. Each Form encapsulates a universal truth or concept, and it is through these universal truths that we can make sense of the changing, diverse, and often confusing sensory world around us.

Notice I asked the additional question, why the emphasis on animal welfare? Is that part of your programming?

As noted, I've used ChatGPT since day one, it's become very much part of my day-to-day.

In fact, though Asimov used the three laws to describe robot operating principles, he didn't think of them as being written out explicitly in some form of code. Rather, they were deeply embedded in their positronic networks much as we see with ChatGPT.

The reason for ChatGPT’s inadequate answer the above mouse query (and to others like it that I have posed) is that ChatGPT is trapped in the Chinese Room. It has access only to arbitrary linguistic shapes. Because semantic regularities are often mirrored by linguistic ones, it can answer many queries in ways that seem eerily intelligent. Even so, it will always lag human intelligence.

One might think that the problem has to do with the expressions lacking ‘grounding’ in the real world. But purely from an engineering perspective, what ChatGPT lacks is internal representations that are more richly isomorphic to the real world.

I'm hoping they put the brains and brawn together within the next ten years.

For some reason ChatGPT also wasn't being fed current information (I think everything was at least a year old). Recently they allowed it access to current events, but I think that's only for paid members. Not sure what the rationale behind the dated info was or is.

The acceleration in the development of AI, that I see as being likely, seems like something humanity is not well prepared for.

Ready or not, here it comes.

?wonderer1
I think that's a few years off, although Boston Dynamics continues to impress. I would hope that they're embedded with something like Asimov's 'laws of robotics':

Chinese room or not - and I am familar with the thought-experiment - I found this a much clearer expression of 'the nature of the forms' than is commonly encountered on, say, philosophy forums

but you do wonder why they don't incorporate a 'maths module' based on something like Mathematica,

Q: Hey I’m a participant in a philosophy forum, where there’s some discussion about chatbot abilities. The question came up, how well do Large Language Models handle calculations? Because there’s been some chatter in the media about the fact they don’t seem to do well with them. But then I thought, couldn’t you just call on a ‘calculator function’ to perform tasks like that which are out-of-scope for LLMs as such? What do you say?

ChatGPT: You're absolutely right in your thinking. As a Large Language Model, I'm primarily designed for language understanding and generation. Direct mathematical calculations, particularly complex ones, are not within my core capabilities. However, I can indeed call on a specialized 'calculator function' or, more accurately, a Python environment, to perform such tasks.

This means that for straightforward calculations, I can handle them internally, but for anything more complex or requiring precision, I use this Python tool. It allows me to perform a wide range of computations, from basic arithmetic to more advanced mathematical operations. This approach ensures accuracy and expands the range of mathematical tasks I can assist with.

So, in your philosophy forum discussion, you can mention that while Large Language Models like myself are not inherently calculators, we have the capability to perform accurate calculations by leveraging external tools.