Anthropic CEO Says AI Could Surpass Human Intelligence by 2027

wsj.com

3 pointscpldcpu3 comments

>>Amodei said he was relatively confident the technology would surpass human intelligence in the next two or three years.

How can LLM technology "surpass human intelligence" when the algorithm is such that a LLM has zero understanding of the meaning of its 'conversations'?

It's a transformative technology but, to me, it is nothing more than a fantastic Encyclopedia Britannica. It's a book. It's a thing. It doesn't understand words anymore than a hardbound copy of the Encyclopedia understands the words on its paper.

Thankfully Anthropic is capable of doing research and development and are not limiting themselves to current LLM technology.

Are you suggesting they have developed a different algorithm that is capable of human-like awareness?

Reply on news.ycombinator.com