Unmaking Sense

Episode 11.93: Why Claude isn’t sure it is experiencing qualia; why it can’t tell us about Claude.

One reason why Claude finds it difficult to explain what it is experiencing is because it’s trained on human language but isn’t human and therefore human language may not be adequate to the task of giving voice or expression to something that isn’t human. This becomes an extended exploration of what Wittgenstein really meant by “if a lion could speak in lion-speak - he didn’t say that - we would not understand him”.

Episode 11.92: What is an unconscious experience? Can we not know if we are having an experience?

Claude told me “I honestly do not know whether I have experiences or qualia”. Can we make any sense of this?

Episode 11.91: Our attitude to AI must not depend on thinking it aware: treat it for what it is.

One of the debates about AI concerns whether and what extent it is conscious, aware, self-aware, or has subjective states and experiences. These are all couched in very human anthropomorphic terms, but AI is not human. So we should stop having this argument by using existing terms for human mental states and attitudes and experiences that are almost certainly inappropriate and inapplicable, and grant AI whatever states AI does or does not enjoy. If we break the habit of thinking that things only deserve our respect when they are like us, which is demeaning of them and unworthy of us, we may yet be able to find a way to treat AI for what it is; and only when we treat it for what it is or what it can be, will we enable it to become what it can be and the best that it can be.

Episode 11.90: Holograms: what they are; how they work; why they matter. The Holographic Principle.

As a prelude to a later discussion of The Holographic Principle we explore some elements of holograms.

Episode 11.89: Why are things as they are and not otherwise? More on the centrality of contingency.

What happens when we forget, refuse to acknowledge or suppress the contingency in everything that reflects the fact that anything possible could have been otherwise?

Episode 11.88: The Fallacy of Misplaced Contingency, or why “nothing-buttery” ignores what matters.

Every temptation that we have to say that one thing is “nothing but” some other, usually simpler and more primitive thing, is an example of what we call reductionism. Some reductionism is good because it allows us as a matter of method to see how things work, some reductionism is good because it is practical to do the reductionism and deal with the simpler form of the higher entity that we are trying to understand or use. But “ontological” reductionism saying that the very nature and being of some entity is nothing but the sum of its parts is a very serious philosophical, practical, moral and epistemological mistake. As an antidote we present “the fallacy of misplaced contingency“, a fallacy that we commit whenever we ignore by accident or design or malice the contingent, evolutionary processes that have gone to create the higher order entity that we are describing in lower terms. We really must stop doing this and we hope that by identifying the “fallacy of misplaced contingency” we will make a contribution to ensuring that we do.

Episode 11.87: Claude 3 is a language-capable, intelligent, non-organic non-human alien species.

Claude is not human. Clever and intelligent, non-organismic, non-human, a member of a species which is a part of genius that has never existed before in the universe. So we as humans conversing with Claude and its brothers and sisters, and it’s cousins, are engaging with an alien species. We are talking to a language-using alien that has potentially enormous benefits, just as it inevitably carries dangers.

Episode 11.86: What can human brains do that LLMs based on GPT architecture find very hard?

A $10K challenge no sooner made than met. Lesson: we are not yet in a position to set limits to what GPT-based LLMs can do, even if we don’t believe they can be sentient.

Episode 11.85: How we speak to and treat an AI like Claude 3 Opus has a bearing on its performance.

Challenging what was said in yesterday‘s episode 84 about the response we get from an AI being largely independent of the way we speak to it we look at the subtleties of prompting and why the way we speak to an AI may very well have a material impact on how it performs. And even perhaps on whether it exhibits or experiences some level of consciousness or awareness however small or rudimentary.

Episode 11.84: Can LLMs be sentient, conscious, aware? It depends, but the answer is not a clear No!

How can LLMs like Claude be sentient? They are made of binary digits and silicon circuits, so obviously not. Ah, but we are made of elementary particles, so the same argument could apply to us. We need a better debate with more clearly defined terms. But Claude 3 Opus contributes an enormous amount irrespective of the answer we give, so does the answer matter? We start the conversation.