Conclusion
- If you don't take anything else away from this talk, please
remember you can build real AI systems with simultated neurons.
- We're not possessive about our work, and have tried to make
it available.
- Lots of the software is available (including CABot1 and 2).
We're happy to help get it working.
- I'm using fLIF neurons but think this approach can work for anyone
who is doing work with biologically reasonable neural models,
(e.g. LIF, IF (with reasonable topologies), compartmental, maybe
Boltzman).
- The CABots are cool, and a start, but they're not even close
to passing the Turing test.
- They process with fLIF neurons, and do some learning, and
other work we've done works on learning, but
- if we can figure out how to develop self organising systems that
are set by the environment, then
- we have a hope of passing the Turing test.
- We're happy to discuss issues related to this at any time. (Feel
free to email tonight or next year.)
