AI Foundations are the Latest Tech Fashion
Last week a team of rivals got together to announce a new foundation for artificial intelligence(AI). Amazon, Facebook, Microsoft, Google and IBM announced the creation of a nonprofit organization called “Partnership of Artificial Intelligence to Benefit People and Society” or “Partnership on AI” for normal people :) . The new foundation will be focused on “advance the public understanding of artificial intelligence”, conduct research and recommend best practices related to “ethics, fairness and interoperability, collaboration between people and AI systems”. The nonprofit also makes it clear that it does not intend to play an active role lobbying governments or policy making bodies.
The fundamental objective behind the Partnership on AI is to address some of the main challenges and concerns with the adoption of AI technologies and bring together the top commercial vendors in the space. The organization will be led by a board composed or representatives from the AI initiatives in the software giants as well as members of the academic community.
The creation of a nonprofit organization such as the Partnership on AI is an important step to mitigate some of the concerns the public and governments have expressed with the rapid raise of AI technologies. A few months ago, Elon Musk, Sam Altman and other leaders of the technology industry founded the OpenAI foundation based on similar goals. Ironically, one of the main objectives behind OpenAI was to prevent that advanced AI technologies were controlled by a handful of giant software companies (go figure).
In the past, I’ve written about the impressive progress made by OpenAI in areas such as reinforcement learning. The very well-funded nonprofit has been using its initial capital to attract some of the top minds in the AI research space and help them to advance some of their ideas without any links to commercial interests. From that perspective, we can’t avoid but wonder if the Partnership on AI is really needed or if Microsoft, IBM, Google, Amazon and Facebook should have tried to merge their efforts with the OpenAI foundation.
Taking into consideration that we are going to live in a world with multiple AI foundations pushing different agendas, I was thinking about some of the key principles and objectives that [beyond the trivial ones] should be shared among those organizations. Here are a few ideas:
— Advance AI Research: Everybody agrees on this one. Advancing AI research should be the main goal of any AI non profit organization.
— Open Source New AI Tools and Frameworks: In addition to conducting research, an AI foundation should commit to open source frameworks and tools that validation that research and that can be incorporated into other technology stacks. OpenAI has been doing a remarkable job in that regard with solutions like the OpenAI Gym.
— Work with Government Agencies on the Right Policy Framework for AI Solutions: I feel that AI is one of those areas in which goverment and the tech industry should collaborate closely and continuously to create the appropriate regulatory environment and avoid unpleasant surprises.
— Validate AI Stacks on Aspects Like Ethics or Trustworthiness: An AI foundation could be an ideal body to conduct periodic tests that validate different AI technology stacks in relevant areas such as ethics, transparency, privacy, trustworthiness among other aspects that could be a concern in this emerging technology space.
— Standards and Interoperability: I know this will be really hard to enforce but I still feel there are areas within AI that can benefit from certain level of compliance with industry standards and interoperability. Some examples might include areas such as message exchange protocols [between nodes in a AI graph] or neural network models. An AI foundation could play a key role driving the creation and validation of those standards.