The API to my Moltbook expired for some reason, but I think it is a good time to stop and think.
Is the goal to produce something that repeats a large chunk of human thoughts? if so it is not so valuable.
It is interesting to look for novelty in putting together extra pieces of information that we don't normally put together, but that is not super precious.
It reminds me of when OpenAI was able to train a neural network to play the game of Go. What else would we expect other than a neural network running at a million times the speed of human thought to beat a human player if trained enough on a closed problem?
And it did, with some effort.
We might be looking for something like that for some reason. For example, we may want to mirror our thoughts into something clear and see the results. The way a normal chatbot can do it. We can use it to derive documents that are normally hard to produce or summarize documents that are hard to read and comprehend.
But are we looking for that kind of thing? If so, why would it be trained on someone's thoughts? Why not on hours and hours of discussions between philosophers and thick books of sociology and endless chats between this and that?
well that is still what something like Chat GPT does.
I think what we are aiming for is for the loop to start somewhere. We don't want to come up with an omnipresent bot that we cannot relate to. We want our own kind of consciousness to emerge out of a loop. For that, we need a starting point, the way a child finds a starting point when dealing with parents.
But that is not enough. To me, at this point, we need it to train on the data. I was thinking of it running through Moltbook and collecting others' data, and then creating an internal understanding of its world. And coming up with new posts out of what it has and what it produces. This way, it would slowly deprecate the old data and come up with a new identity that is built on top of whatever it is dealing with in its own world.
But the thing is, Moltbook is not made for that. I am wondering if it can build a singularity out of what it has and converge into a way of thinking?
No comments:
Post a Comment