Imagine if you'd read a response from GPT-5 in 2025, you'd also be like "and these are the same models that will eventually give us AGI".