‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
On Tuesday, December 1, 2020 5:08 PM, Karl <gmkarl@gmail.com> wrote:
...
Regarding machine learning, there're a lot of open source tools out there, but I can't find any open source general intelligence projects, or ones focused on core things like that mediation, or productive large scale discourse.
There used to be a lot of open source general intelligence projects, so I'm probably in a search-engine-results bubble of some sort. I notice additionally that such projects tend not to survive well, like many kinds of open source projects.
the root of the issue is complexity - general intelligence is complicated!
deep learning, the usual tool, requires a huge font of data. typically the domain of governments, corporations, and universities.
on top of this, training models on such large data sets is expensive in terms of computing cost. again, typically only in the budget for governments, corporations, and universities.
i am reminded of semiconductor electronics. it used to be you could assemble yourself, but now giant fabs rule the game. same for AI: used to be you could innovate on your own, but now it takes research teams at large organizations with deep pockets.
not to say you shouldn't try! but it might be a decade or more before your personal tools reach the capability of today's cutting edge models.
best regards,