Current LLMs Lack Consciousness, Not Advanced AI Capabilities

Key Quotes

"Weinstein is right in how he starts the conversation right he's saying look we can't just saying uh language models predict the next word uh that's gar that's too dismissive so while it's true that yeah they literally do predict the next word there is a lot of impressive understanding and processing that goes into actually figuring out what word to to to produce in fact right in the new yorker recently james summer argues that we can even think of the processing that goes into predicting next words in language models as thinking and i think he lays out a good argument for that"

Cal Newport agrees with Brett Weinstein's initial point that dismissing language models as merely predicting the next word is too simplistic. Newport highlights that significant understanding and processing are involved in this prediction. He also notes that James Summer's argument, suggesting this predictive process can be considered a form of thinking, is a valid perspective.

"No way okay what we actually have is something so analogous to a child that that is the right model in other words when a baby is born it has no language it may have some structures that language will slot into but it doesn't have any language it is exposed to tons of language in its environment it notices patterns right not consciously notices but it notices them in some regard you know that every time somebody says the word 'door' you know there's a fair fraction of those times that somebody you know opens that portal in the wall i wonder if 'door' and that 'portal in the wall' are connected whatever it is so the quaint is a child goes in a matter of a few years from not being able to make a single articulate uh noise to being able to speak in sentences make requests to talk about abstract things that is an llm"

Newport acknowledges that Weinstein's analogy of language models to a child learning language has some functional alignment. Newport explains that both babies and LLMs learn by exposure to vast amounts of language, identifying patterns and connecting words to concepts. This functional similarity, where they build an understanding of linguistic concepts, is where Newport finds common ground with Weinstein's observation.

"All right now weinstein has lost me he has moved from describing uh how these appear to behave on the outside to trying to describe what's happening inside the llm and the way he's describing is not how a language model actually operates language models don't run experiments in the way a human mind might to help it better figure out what's going to what to do or to see what will happen llms certainly don't want anything to happen they have no values or drives like a human brain does"

Newport states that Weinstein's argument becomes problematic when he attempts to describe the internal workings of an LLM. Newport clarifies that language models do not operate like a human mind by running experiments or having desires, values, or drives. He emphasizes that this shift from external behavior to internal processes is where their understanding diverges.

"So what weinstein is talking about is actually the way a human brain works but llms are way more static and sequential and simpler to describe their operation than a human brain"

Newport draws a clear distinction between human brain function and LLM operation. He asserts that Weinstein's descriptions align with how a human brain operates, characterized by dynamic and spontaneous processes. In contrast, Newport characterizes LLMs as significantly more static, sequential, and simpler in their operational mechanics.

"Regardless of the complexity of the information and logics that is represented by those numbers that static process where nothing changes it's non spontaneous it just moves sequentially through layers and produces a word on the other end does not and is incapable of matching any reasonable definition of consciousness"

Newport argues that the static and sequential nature of an LLM's operation, even when processing complex information, fundamentally prevents it from achieving consciousness. He explains that the process involves a fixed set of numbers and a predictable algorithm that moves through layers to produce an output. Newport contends that this mechanical, non-spontaneous process cannot align with any reasonable definition of consciousness.

"The scientist tries to figure out the actual underlying mechanisms behind something they're observing and then once they understand those mechanisms they can use that to reason about what else might be possible or what else might happen"

Newport contrasts a scientific approach with what he terms an "ancient animist religion" approach to understanding the world. He explains that scientists focus on uncovering the actual underlying mechanisms of a phenomenon. Newport states that this mechanistic understanding then allows for more accurate reasoning about future possibilities and outcomes.