Considering that these AI tools are built off models that were trained with large datasets, we can say that the "clean slate" we all start with is not so clean, it's memory in disguise.
We don't just draw from this memory as we engage with these tools, we also (re)write new things into it, adding new layers as we interact.
I enjoyed reading this.
Considering that these AI tools are built off models that were trained with large datasets, we can say that the "clean slate" we all start with is not so clean, it's memory in disguise.
We don't just draw from this memory as we engage with these tools, we also (re)write new things into it, adding new layers as we interact.
i really resonate with this
memory is essentially the next step in enhancing communication as I see it as mapping knowledge/information to a user’s world model
the best teachers / communicators do this
i also wonder if the MCP layer will have its own memory and how “memory to memory” interaction will look like
My gut says they’ll be stateless but be the layers where memories are transferred.
Refreshing read. This will be life saver in building discipline. However this is gonna shape our relations in a way that we would not ideally follow