We come to bury ChatGPT, not to praise it

>> You’re just repeating that computing is not predicting without explaining why.

Yes, because I assume you're a programmer, or a computer scientist, and you already understand the difference.

To clarify, the difference is that a program that computes the most likely next token in a sequence, like a language model (LM) is one kind of program, whereas the set of all program contains infinitely many programs that are not LLMs; and an LLM cannot calculate every such program, nor can it perform the same calculations as those programs, it can only calculate the most likely next token in a sequence.

Is the distinction clearer now? Also, are you a programmer? If so, I would point to the wikipedia article on Turing Machines or the one on the theory of computation. It's perhaps not a very easy subject to explain in a HN comment, but if you have some background in programming computers you should be able to grok it. You will see for example that a Turing machine is a device that manipulates symbols to produce other symbols, which is what we generally mean by "computation" and not a device that predicts the next token in a sequence, as an LLM does.

Sorry if I assume too much about your background - this is HN after all.

I run the cheaper self hostable OpenAI alternative https://text-generator.io I've been working on automating this manual verification of everything, with a few components we already have like a search engine and an edit API we can both detect and correct most of these errors to at least be reflective of what a reliable source says like Wikipedia, still a lot of reasoning, logic and math issues will remain, but there's a big step up coming soon in factual generation

Picture a line—say, the line defined by y=2x + 3.

Now, tell me what's contained inside that line. Not "what does it mean" or "what's it made up of"; "what is contained inside it?"

The question doesn't make sense. There's no "inside" of a line. It's a one-dimensional mathematical construct. It fundamentally cannot "contain" anything.

"What's inside the line?" is a similar question to "Is ChatGPT self-aware?"—or, more aptly, to "What is ChatGPT thinking?" It's a static mathematical construct, and thinking is an active process. ChatGPT fundamentally cannot be said to be thinking, experiencing, or doing any of the other things that would be prerequisites for self-awareness, however you want to define that admittedly somewhat nebulous concept. Thus, to even ask the question "Why don't you think ChatGPT is self-aware?" doesn't make sense. It's not that far different from asking asking "Why don't you think your keyboard/pencil/coffee mug is self-aware?"

The intelligence of all humans is roughly analogous in ability—even if a given human has not learned to do formal logical deduction and inference, the fundamental structure and processing of the human brain is unquestionably capable of it, and most humans do so informally with no training at all.

Attempting to cast doubt on the human ability to reason, to comprehend, and to synthesize information beyond mere stochastic prediction reflects a very naïve, surface-level view of humans and cognition, and one that has no grounding in modern psychology or neuroscience. Your continued insistence, through several sub-threads, that we cannot be sure we are any better than ChatGPT is very much an extraordinary claim, and you have provided no evidence to support it beyond "I can't imagine a proof that we are not."

Maybe go do some research on how our brains actually work, and then come back and tell us if you still think we're all just predictive chatbots.

Try https://text-generator.io the bs generated is about the same tbh... But much cheaper, and combines a web crawler so it can speak about links and images which makes it a lot better at lots of things like making a believable conversation about designs/invoices/reciepts/emojis.

[dead]