I have been thinking about this exact topic recently and would be happy to engage in dialog with you about it. Perhaps it would be fruitful to inquire into the API(s) of the brain to drive development of software which uses analogous API(s)? Perhaps cell walls can be thought of as an API. The walls only allow certain things into the cell while blocking others. And inside the cell you have DNA which encodes for how things (that are in the cell) are to be transformed, much as methods transform their inputs. Is the cell the appropriate level of abstraction for AGI? Perhaps we are more interested in the API's between two neurons (synapses)? Or perhaps we are interested in understanding how large collections of neurons interface with other large collections of neurons (nerves, commissures, etc.)?
It certainly is the case that in the brain certain areas of the brain perform specialized transformations on kinds of information (visual cortex, audio cortex, motor cortex, etc.) And so certainly we can say (at least at a high level) that these regions of the brain have API's such that they accept information of those associated kinds. Of course it isn't like the brain has a sense of class types as we do in programming languages, rather it simply has neurons. As such a module that does visual processing if (for what ever reason) were connected to the audio cortex, then you would begin to hear those things you see. Do we want our API to be such that we allow for this kind of behavior? Do we care about biological plausibility? Or do we simply care about creating something that works? I think one problem with AGI designs is that as our understanding of the brain increases our implementations often need to be reworked in order to take into account the new understandings. Is it possible to future proof ourselves from this? On Sat, Mar 14, 2015 at 5:02 PM, Steve Richfield via AGI <a...@listbox.com> wrote: > IMHO the future of ALL AGI approaches lies in the careful design of the > APIs and other interfaces that glue it all together. I have in the past > called for a group effort to rough something out - and been met with > deafening silence. > > Perhaps there is wisdom in the KnowmAPI *specifications* (vs. its > implementation) to start such a discussion? > > Steve > ==================== > > On Sat, Mar 14, 2015 at 5:38 AM, <a...@listbox.com> wrote: > >> Any thoughts on that? >> >> http://www.knowm.org/ >> >> >> >> >> http://www.reddit.com/r/artificial/comments/2yczb6/ktram_a_memristorbased_machine_learning/ >> >> >> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >> <https://www.listbox.com/member/archive/rss/303/10443978-6f4c28ac> | >> Modify <https://www.listbox.com/member/?&;> Your Subscription >> <http://www.listbox.com> >> > > > > -- > Full employment can be had with the stoke of a pen. Simply institute a six > hour workday. That will easily create enough new jobs to bring back full > employment. > > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/26973278-698fd9ee> | > Modify > <https://www.listbox.com/member/?&;> > Your Subscription <http://www.listbox.com> > ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
