AGI is a human level machine intelligence. Think about all the sci-fi movies and games with robots that can talk and understand humans perfectly, and whether th

AGI is a human level machine intelligence. Think about all the sci-fi movies and games with robots that can talk and understand humans perfectly, and whether they are consciousness or not is a separate question. A good example is the Gith from Mass Effect that have an entire civilization (if that's what we could call it), but their intelligence is a different and behave more like a hive mind. Another example could be C3PO or even R2D2 (given that they both understand human language, intentions, etc). > I mean surely a calculator can pass human levels of computing speed and accuracy on multiplying large numbers. You can't talk to calculators, something that humans do. Transformers (GPT, Claude) are much closer to AI that we can talk to, however it becomes evident they do not think like humans despite their impressive amount of knowledge. I would say the biggest missing piece of the puzzle with LLMs and transformers is theory of mind, or the capacity of humans to model the thinking of other humans. Once AI can model your thinking base on the context of the conversation, it will be capable of holding long conversations and provide useful, realtime insights. I think at that point we will have AGI, and we will still not have a full understanding of consciousness or whether AGI is conscious.