WBAI Policy, FY2019


Four years have passed since 2015 when a number of AGI-related organizations (including ours) were established.  In these years, we have seen increasing AGI-oriented problem settings across multiple task domains in the machine learning research community, such as multitask learning, continual learning, and lifelong learning.  As the R&D of AGI progresses and is recognized widely, its impact (whether being good or bad) on the humanity is recognized as well. Thus, it will be difficult for an organization that pursues the R&D of AGI to gain trust, unless it democratizes the technology and develops AGI in a beneficial form.

Under these circumstances, we decided to organize the knowledge in the neuroscientific community into specifications for AGI development and using it to promote democratic collaborative development in the AI​​/ML community.  Namely, we shall serve as a bridge between the neuroscientific community and the AI​​/ML community. The development methodology for brain-inspired AGI was formed in 2018 with our experiences in hackathons over years. With this background, we plan to conduct the following educational and R&D businesses for FY2019.

Education business (human resource development)

The objective of the education business is to develop long-term interdisciplinary human resources with different specialties such as AI/ML, neuroscience, and cognitive science necessary for R&D in the WBA approach.  We will continue our regular activities by holding seminars and an annual symposium to raise public interest.

This year, we support the Animal-AI Olympics organized by CFI, instead of holding the annual hackathon.  We sponsor the WBA award for the Animal-AI Olympics for the most biologically plausible entry. By holding related events in Japan, we aim for increasing participants interested in WBA.

We shall participate in  academic events and cooperate with and academic organizations in Japan and overseas.

We shall cooperate with external groups with regard to human resource development and find new people to be involved in the WBA approach.

We shall issue the WBAI Incentive Award to praise those who have achieved significant results in brain-inspired AGI development.

R&D business

The objective of the R&D business is to accelerate democratic AGI research by promoting external research through the WBA approach.  We promote activities not to compete with existing research institutes, and revitalize research by, e.g., implementing machine learning models with open platforms.

In FY2019, we will concentrate on compiling neuroscience knowledge and developing it as useful information for brain-inspired AI development.  This is an important step toward the construction of the Whole Brain Reference Architecture (WBRA) as data for brain-inspired AGI design.  Specifically, we plan to design functional circuits as part of WBRA, an ontology for describing them, and an environment for storing the design.  In terms of functional circuit design, the basal ganglia framework will be designed following the neocortex and hippocampus.

We shall continue to promote the development of integrated software platforms for building the WBA.

We shall develop models for aphasia and other cognitive impairments while relating them with our other R&D activities.

We plan to continue activities to make the brain-inspired AGI more beneficial.

Most of these R&D activities will be carried out with research funds from external research institutes.  In FY2019, we will have our own research activities as we took over some of researches conducted at Dwango AI Lab.

Background of the Policy

When we founded WBAI four years ago, we declared:

We establish the Whole Brain Architecture Initiative as a non-profit organization that supports R&D based on the WBA approach over the long term, aiming at the early realization of artificial general intelligence.

After that, while the advent of AGI is approaching, it is not easy to obtain supports for AGI development that is not yet in the practical stage as the third AI boom settles down.  As an NPO that does not share profits, WBAI will face a financial difficulty for years to come.

Recently, it was reported that Microsoft invested $ 1 billion in OpenAI LP, a subsidiary of OpenAI, a non-profit organization that advocates the development of beneficial AGI, and became the preferred partner for commercialization of AI technology created by OpenAI (June 22, 2019).  The news caused wide repercussions. As OpenAI’s position is close to us, its outcome will be a lesson for us.

Let’s take a look at our objectives at its inauguration in August 2015:

  • Sustainability: retaining its objectives towards an expected advent of AGI (beyond 2030).
  • Publicness: publishing research results and developing software with an open license.
  • Cooperation with academia: promoting interactions among academic areas such as neuroscience, cognitive science and machine learning, with AI as the pivotal area.
  • Human resource development: fostering multi-domain knowledge required for WBAI research.
  • Infrastructure research: creating platforms for machine learners, evaluation methods for AGI, and research environments for simulation and data preparation.
  • Social activity: making advanced AI technologies transparent for the public to create the future with AI.

As the footsteps of AGI approach, a few international organizations that advocate avoiding the monopoly/oligopoly of the technology and promote AGI development for the public good over the long term will be valuable.

Thus, we are aiming for a vision of a world with AI in harmony with humankind, and would like to keep moving forward without forgetting our original objectives.  From FY2019, we shall proceed with the following steps to realize our objectives.

  • Promote the accumulation of design data for brain-inspired AGI while refining development methodology
  • Develop active human resources and continue holding events such as seminars to appeal our objectives
  • Examining and publicizing the impact of AGI (including WBA) on society

We will steadily improve the development methodology and environments for brain-inspired AGI, and continue our efforts to create an open larger-scale democratic community for WBA development.