I wonder if the argument over open source vs proprietary is a little too simple. Instead the model should be more like Lucene/Solr is for Search/Information retrieval where there is a core open source architecture but much of the algorithms and processing extending that architecture is done by proprietary companies. There's a relationship then between open source and proprietary that is mutually beneficial. When it makes sense to push expansions from companies back to core then it can be done so, to the benefit of the larger community. The open source community can focus on maintaining and expanding the core architecture and functionality, while companies can still make money off of applications that they develop.
It seems like for AGI-ish work, it'd be beneficial to have that type model where there's very simple APIs that can be used by individuals for specific applications, like game-AI or NLP. The core APIs can then be extended for proprietary applications, or if it maks sense then those extensions can be pushed back to the core open source community. Not trying to criticize OpenCog. I've read most of the first draft of BBMand largely agree with the overall ideas. It's just so hard to use and get introduced to. There's no straightforward API that others can use to build applications and easily get into the system. -Chris On Saturday, March 22, 2014 6:13 AM, Ben Goertzel <[email protected]> wrote: Hey Azn, Of course I would take US$100M or even US$20M personally, even if it required me to work for the company that paid the $$ for 3 years or whatever [a common sort of requirement] ... However, I don't currently have the sort of company that a Google or Facebook would want to buy.... Deep Mind had ~60 machine learning programmer/scientists, colocated, with fancy pedigrees -- and they put a lot of work into making working demos that would impress folks like Google execs (e.g. video game playing by reinforcement learning & computer vision, which is not fundamentally hard but requires lots and lots of fiddling to be surely).... And they worked up to that acquisition via taking $$ from Silicon Valley insiders (Peter Thiel's Founders Fund) and cultivating associated relationships.. Vicarious Systems has a similar, though far smaller team, and also had Founders Fund $$; and also put a lot of work into making the right kind of working demo .. in their case solving the CAPTCHA problem.... Again, solving CAPTCHA is not incredibly hard given a deep learning computer vision system, but it's a long road of fiddling and tweaking given current technology... Neither company had/has a great near-term biz model, but both were well primed for tech biz acquisitions Point is, grooming a company for potential acquisition or investment by Silicon Valley types is a specific quest that takes time and special effort. It's not a bad path to take. But it's not the case that these companies just want to buy any high-quality AI team (and I emphasize the team because IMNSHO Deep Mind was mainly an acqui-hire); they tend to work thru their own social networks; etc. Since I don't have a company of the sort these firms like to buy, my opportunity regarding such firms would be to get one of them to make a job for me, and a suitable fraction of my OpenCog colleagues, working on OpenCog type AGI software. Most of these companies like proprietary code so they would probably want us to make a proprietary fork of OpenCog and integrate with their own internal software systems. While there would be no multimillion dollar hiring bonus, there would be the option for a large personal $$ payout if we made our software do something of value to that firm a few years down the road.... And of course there would be the option to convince that big company to progressively put more and more $$ into said proprietary OpenCog fork... If I didn't have other interesting options available that might be an appealing path for me and some of my colleagues. It's not impossible I would take that route. Yet I tend to be a free spirit; I like to be able to work from home and to travel freely, and to talk freely about what I do.... And I have a strong intuition that it's better to make AGI in the open source way, drawing insight from a wide group of folks around the world with various backgrounds... If my work with Aidyia Limited, a HK hedge fund I've co-founded, goes well, then I can make enough $$ myself to fund OpenCog at a level that will eliminate the need for megacorporate sponsorship. But of course financial prediction is risky even with machine learning algorithms on your side, and work on that application takes time that could be spent on AGI... And so it goes in early 2014... ;) Ben On Sat, Mar 22, 2014 at 1:50 PM, Azn A <[email protected]> wrote: Interesting, but how is selling some of your low level work forcing you into a megacorporation? You can probably walk away once the deal is done right? With what, $100 million? You say you would only need $60 million to get an AGI.. > > > >On Fri, Mar 21, 2014 at 10:38 PM, Ben Goertzel <[email protected]> wrote: > > >>Eventually there will be an AGI that is far beyond human control. But before >>that, there will be a gradual path of increasing AGI, when the relationship >>btw humans and AGIs will be more complicated. Having this >>intermediate-stage AGI guided primarily via corporate motives of short-term >>profits, seems unwise to me... >> >> >>Also, it seems better if all the smart minds in the world can collaborate on >>creating AGI, not just the few who happen to have been hired by some >>particular company >> >> >>Joel Pitt and I made these arguments in more detail at >> >>http://jetpress.org/v22/goertzel-pitt.htm >> >> >> Linux demonstrates that OSS can go as far and fast as in-house projects at >>big companies (or more so), if things fall into place right... >> >> >>-- Ben >> >> >> >> >> >>On Sat, Mar 22, 2014 at 1:33 PM, Azn A <[email protected]> wrote: >> >>Ben Goertzel wrote: >>>> I would prefer to make AGI open source, not within a megacorporation.. >>> >>> >>> >>> >>>Why? It's like not the AGI can be controlled or whatever... >>> >>> >>> >>>On Fri, Mar 21, 2014 at 10:28 PM, Alan Grimes <[email protected]> wrote: >>> >>>Ben Goertzel wrote: >>>>> I would prefer to make AGI open source, not within a megacorporation... >>>> >>>>Got detailed build instructions for Linux? I really need to know what >>>>the status of your project is right now and my only usable computer is a >>>>linux machine. >>>> >>>>-- >>>>IQ is a measure of how stupid you feel. >>>> >>>>Powers are not rights. >>>> >>>> >>>> >>>>------------------------------------------- >>>>AGI >>>>Archives: https://www.listbox.com/member/archive/303/=now >>>> RSS Feed: https://www.listbox.com/member/archive/rss/303/10514698-9a8cda1e >>>> Modify Your Subscription: https://www.listbox.com/member/?&; >>>> >>>>Powered by Listbox: http://www.listbox.com >>>> >>> >>>AGI | Archives | Modify Your Subscription >> >> >> >>-- >>Ben Goertzel, PhD >>http://goertzel.org >> >>"In an insane world, the sane man must appear to be insane". -- Capt. James >>T. Kirk >> >>"Emancipate yourself from mental slavery / None but ourselves can free our >>minds" -- Robert Nesta Marley >>AGI | Archives | Modify Your Subscription > >AGI | Archives | Modify Your Subscription -- Ben Goertzel, PhD http://goertzel.org "In an insane world, the sane man must appear to be insane". -- Capt. James T. Kirk "Emancipate yourself from mental slavery / None but ourselves can free our minds" -- Robert Nesta Marley AGI | Archives | Modify Your Subscription ------------------------------------------- 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
