imagine coming back to a conversation with chatgpt from 2023.
it wouldn't even question it.
models live in this discrete timeline.
each time you send a prompt, the agent starts operating
until it finishes its task and waits until the next prompt.
in between one prompt and another is nothing, at least for the model.
while yes, we do provide the model with time-stamps, what if this isn't the only thing it needs?
what if our continuous relationship with time provides us with something unique?
something that models today don't have: experience.
for us, experience shapes who we are
and who we are shapes what our new experiences will become.
think of it this way: a neurosurgeon with 20 years of experience and an NBA player with 20 years of experience can walk into the same room and experience two completely different things.
even though the environment is the same
the experience isn't.
the 20 years of divergent experience shapes what their attention focuses on.
this attention shift is what shapes the experiences they have.
yet paradoxically, these very experiences are what make up the person in the first place.
the intricate dance between how you shape your environment and how it shapes you might be part of what makes us feel human.
we can't be certain until we actually achieve fully continuous agents who don't catastrophically forget.
but maybe once we do, then we'll really see the differences (or similarities) between human and artificial intelligence.