On Mon, Nov 2, 2015 at 8:44 AM, Steve Richfield <[email protected]>
wrote:

> Ben,
>
> Granting everything you posted here, it seems that the world falls cleanly
> into two camps:
>
> 1.  Those who are doing research with no plans or funding leading to
> substantial profit. OSS works well in this paradigm.
> 2.  Those developing products given adequate financing, so they have money
> to spend, are NOT willing to share their IP without receiving BIG payments,
> etc. Here, trade in pieces seems to be the best paradigm.
>
>

No, the world today is not so clean as that...

Developing OSS AGI code is still consistent with building profitable
businesses that use this code (coupling it with proprietary code) to
provide services to customers in vertical markets....  Ever hear of
RedHat?  It's not Microsoft but it's doing OK...   Other examples:

https://www.quora.com/Who-are-the-most-successful-open-source-based-companies

..

Anyway, getting from here to AGI seems to me to require *both* solving a
bunch of tricky research problems AND doing some large-scale software
engineering right....

If the OpenCog design is basically right, then still we have maybe 200-300
human-years of effort to get to a human-level AGI, according to my best
attempts at fully realistic estimates....   To do this in a startup company
context requires one to make a viable business using software created
incrementally along the way, and use this interim software to drive revenue
or fundraising to get the whole thing done.   This is very difficult
because one is trying to do two things at once (build an AGI and build a
business).   Some alternatives would seem to be:

-- work within a big company like Google, Baidu, Facebook, IBM, etc., which
have massive resources

-- get a zillionaire to donate lots of $$ (or become a zillionaire, i.e. at
least a tens-of-millions-aire, and self-fund)

-- get massive gov't funding like Markram did for the EU Human Brain project

-- build a Linux-level OSS AGI project

...

I gravitate toward the fourth option because it's more DIY (though
certainly it's not easy either, and we have a number of steps to carry out
to get there w/ OpenCog).  And I note that this option does not rule out
gov't or wealthy donor participation --- nor big company participation
(e.g. Google, IBM, etc. have contributed substantially to Linux)

OTOH, if you believe you have an AGI design that can be brought to the
human level or near with much more modest effort, then you may have a
different menu of options available...

If your goal is to make a lot of $$ then whether OSS is the right approach
is arguable.  It's not a non-workable approach to profit, but neither the
simplest approach.  But if your goal is to make AGI there's a lot to be
said for it, and it certainly doesn't rule out making $$ along the way...

-- Ben G



-------------------------------------------
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

Reply via email to