This is a good argument. While models for producing useful text and images will continue to improve and create evermore convincing outputs, they lack some funda

This is a good argument. While models for producing useful text and images will continue to improve and create evermore convincing outputs, they lack some fundamental aspects of lived experience. Imagine if we could record all (or many) aspects of our daily experience from 1M+ viewpoints over many years - that would lead to a revolutionary model and would be unfathomably expensive to train. The sounds heard waiting for a train. The experience of going on a date or hiking or swimming. The tastes and sensation of biting into some fruit. They each hold such a rich multi-modal experience that feels impossible to replicate in a model at at this time. An LLM can describe it but it cannot experience it. Not that such experience is required to do what an AGI might be asked to do, but how could something reach AGI or higher without that level of experiential detail?