Donate
From OpenCog
A lot of people talk about the possibility of creating advanced AI, with capability at the human level and beyond.
Here at OpenCog, we're actually doing it. We have a detailed plan and the ability to execute it, and we're proceeding with the hard work step by step.
You can support our efforts by making a tax-deductible donation to the OpenCog Fund, which is managed by the Singularity Institute for AI (SIAI).
100% of donation money is used to pay individuals to develop the OpenCog software according to our existing detailed designs, in the direction of advanced, beneficial AGI.
Note our limited special offer: the first 1000 people to donate $100 or more to OpenCog will receive one of 1000 autographed first editions of the forthcoming book Building Better Minds: Engineering Beneficial General Intelligence by Ben Goertzel.
How to Donate
OpenCog donations can be made via credit card, and are processed via the NetworkForGood website, which you reach by clicking on
Note that:
- Donations made to SIAI via the above NetworkForGood link go directly into the OpenCog Fund
- Donations made to SIAI via the "donate" page on SIAI's website go into SIAI's general fund
Also, in the happy event that you are interested to make a relatively large donation (let's define this as US$5000 or more, just for simplicity), then it may be preferable to us a mechanism other than NetworkForGood, for which you may contact Ben Goertzel via ben at goertzel dot org. NetworkForGood, like PayPal, charges a nontrivial credit card processing fee, calculated as a percentage of the donation amount. This is not a problem in itself (online credit card processing is a valuable service), but it is preferable to bypass this fee in the case of more sizeable donations.
Donate $100 and Get one of 1000 Autographed First Edition Copies of "Building Better Minds"
As an incentive for donation, we are offering the first 1000 people to donate $100 or more an autographed first edition copy of the forthcoming book Building Better Minds: Engineering Beneficial General Intelligence.
The first edition is expected to be released in the second quarter of 2010.
To take advantage of this special offer, you can click the button on the following page
http://goertzel.org/BuildingBetterMinds.htm
and donate $100 or more via Google Checkout (which will allow you to specify your mailing address as well as to donate via a credit or debit card).
Alternately, if Google Checkout doesn't work for you, you can also make the same "special offer" pre-order/donation via the Network for Good. In this case, be sure to put the text "Building Better Minds" in the "Designation" field, to specify your address in your Network for Good profile, and to select the option indicating that your address should be shared with the Singularity Institute.
A (minority) portion of the funds raised via these donations will be to pay for part of Dr. Goertzel's time as -- in parallel with carrying out technical OpenCog work -- he completes the book, which is currently in rough-draft form. The rest will be used to pay for OpenCog software development.
Drawing on material from the OpenCog wiki and prior publications, but also introducing significant new material, the book will lay out the basic ideas on which the OpenCog Prime approach to creating beneficial artificial general intelligence is founded. Mathematical and software architecture oriented information will be presented in appendices, so that the main body of the book will be accessible to the educated layperson.
Fall 2009 Fundraising Goal: US$100K
Our goal during Fall 2009 is to raise US$100K for OpenCog, which will allow us to pay a core OpenCog team for the 5-month period November 2009 - March 2010. Spent at $20K/month, this will allow us to fund
- Full-time work by Dr. Nil Geisweiller (Bulgaria), Samir Araujo (Brazil)
- 2/3 time work by Dr. Joel Pitt
- Part-time work by Dr. Ben Goertzel (USA), Cassio Pennachin (Brazil), Welter Silva (Brazil)
Project Goals for Nov 2009 - March 2010 if Fundraising Goal is Achieved
Part of what the OpenCog system does now can be seen by looking at the three "AI in virtual worlds" videos linked from
These videos show OpenCog controlling a virtual agent in an online virtual world, supplying it with the capability to learn by imitation and reinforcement, to interpret simple English, and to use its embodied experience to help it disambiguate language.
However, the OpenCog system contains significant capabilities not demonstrated in these videos, because while they exist in the OpenCog codebase they are not integrated with the virtual-agent-control system.
Our main goals for Nov 2009 - March 2010 if the $100K funding is achieved are to:
- Connect OpenCog's Probabilistic Logic Networks engine to the OpenCog virtual agent control framework, enabling the system to carry out simple reasoning regarding its virtual surroundings and its relationship to them. For instance, this sort of reasoning would allow the agent to figure out what to do with an unknown object, based on the known properties of known objects that are similar to it in various respects.
- Connect OpenCog's NLGen English language generation system to the OpenCog virtual agent control framework, in a way that allows the system to carry out simple dialogues about what it's doing in the virtual world
- Modify the way OpenCog's (evolutionary-algorithm and hillclimbing based) learning engine is connected to the OpenCog virtual agent control framework, so that the agent is able to freely learn from its unsupervised experience. (In the current virtual agent control framework most of the learning occurs based on direct interactions with a teacher, even though the learning algorithms themselves are much broader than this.)
These goals of course do not achieve OpenCog's long-term goal of AGI at the human-level and beyond, but they constitute significant steps forward, building on the work we have done so far.
If more funding is achieved we can progress even further toward our long-term goals during this period; and if partial funding is achieved we can achieve some but not all of these goals.
What Happens After March 2010?
The current fundraising drive is specifically intended to raise funds for the project for the period Nov 2009 - March 2010
Funding beyond that point may be achieved via further fundraising drives, and/or potentially via other mechanisms such as commercial OpenCog-based projects executed by Novamente LLC or other firms.
Government grants for OpenCog work have also been applied for, and if one of these is received then this would partially cover the cost of OpenCog science and engineering.
"One step at a time."
Pledges Regarding Use of Funds
To keep things really simple, we make two pledges regarding the use of funds raised during this fundraising drive:
Pledge 1) 100% of funds received from donations to OpenCog will be used to pay scientists and engineers to do AI software design and development on the OpenCog project, oriented directly toward the creation of beneficial Artificial General Intelligence at the human level and ultimately beyond.
Pledge 2) All paid OpenCog design and development will be done at below-market rates (relative to the developer's location), making use of developers who are highly devoted to the OpenCog mission. Furthermore, we will make all feasible efforts to minimize costs without sacrificing work quality and progress, e.g. by using developers in inexpensive locations.
We are operating with zero overhead expenses: no office, free online resources wherever possible, and donated legal and accounting support. And at the moment it's not hardware that we need, it's human time and brainpower, to turn the existing OpenCog AGI software designs into working, tested, intelligent code.
How Has OpenCog Been Funded So Far?
OpenCog was founded in early 2008 via a collaboration between Novamente LLC and the Singularity Institute for AI. Novamente donated a substantial amount of source-code, as well as the ongoing part-time efforts of a number of staff programmers and scientists. SIAI donated $125K that was used during 2008 and early 2009 to pay Dr. Joel Pitt, Gustavo Gama, Welter Silva and others to develop OpenCog code.
A boost to OpenCog was provided by participation in the 2008 and 2009 Google Summer of Code programs, under which Google funds university students to do summer internships on open-source projects. We had 11 GSoC students in 2008 and 9 in 2009. Each year around half of the student projects were dramatic successes, which is a good ratio. We believe we have reasonable odds of being selected by Google for GSoC 2010 as well.
During the period March-August 2009, Novamente LLC funded Dr. Joel Pitt, Dr. Nil Geisweiller, Welter Silva and Fabricio Silva to work on OpenCog (along with funding Dr. Ben Goertzel, part of whose time was devoted to design and coordination for OpenCog).
At the moment (September 2009) SIAI has shifted its focus away from immediate-term AI software development and is directing its resources toward the ethics and theoretical foundations of advanced AI. And, due to the recession economy, Novamente LLC has been forced to focus more narrowly on its AI consulting business, meaning that it is not currently in a position to fund much OpenCog software development.
Why Does a Free Open-Source Project Need Money?
This is really not a very good question, but it's asked often enough that we feel the need to provide an answer here.
As valuable as unpaid volunteer efforts are for a project like OpenCog, we've found that much more AGI work can get done much faster if we're able to pay some of the right people to work on OpenCog. This is just economic reality: even the most dedicated AGI geeks need food and shelter and computers ... and many have family to support and various other financial obligations.
