Comment by achristmascarl

Whether or not LLMs are conscious, it does now seem possible to build conscious systems with LLMs as the key enabling component.

Many of the goalposts offered here and elsewhere, like embodiment or emotions, can be and have been emulated or approximated and built around an LLM, with the LLM acting as outsourced intelligence (or cognition, or reasoning, or some other better term). Will this system be conscious? Who knows. It certainly won't be human. But I suspect much of the concern around AI being conscious has less to do with whether an LLM itself is conscious and more to do with the fact that we are now capable of building potentially conscious systems, and the limits to what those potentially conscious systems are capable of keep expanding as the capabilities of LLMs grow.

Also, is "Claude" an LLM? I know Claude Opus 4.8 is an LLM, but Claude is simply a label that Anthropic chose for a series of products and services, many of which are LLMs. Anthropic itself describes Claude as "a next-generation AI assistant" [0]. Right now, we use Anthropic's "Claude" LLMs through various channels, but nothing's stopping them from tacking on various systems to try and make Claude a conscious entity.

[0] https://www.anthropic.com/news/introducing-claude