Some of the top technology companies that have already invested heavily in machine learning technologies and services have joined together in a consortium to share research and create best practices on AI technologies. The Partnership on AI’s initial founders include Amazon, Facebook, IBM, Google/DeepMind, and Microsoft.
The consortium will accept other members in the future, including non-profits and activists, and they will get equal membership roles in the group along with the corporations. Other important AI players in the industry who seem to be missing from the group right now are Apple, Baidu, and OpenAI. However, they may eventually become members as well, once membership is open, unless they have some other objections to the goals of the Partnership on AI consortium.
Stated Goals
The group wants to support research and recommend best practices in areas such as ethics, transparency, interoperability, privacy, reliability, and robustness of technology. It also aims to increase awareness and understanding about AI for the public, presumably so the public keeps seeing the rise of AI as a positive evolution, rather than a negative or dangerous one.
Lastly, the group wants to create an open platform where researchers can communicate with each other about new issues or advances in technology.
Mission
The consortium will engage experts not just in machine learning, but also in psychology, philosophy, economics, finance, sociology, public policy, and law, to discuss emerging issues related to AI’s impact on society.
The group also wants to finance objective third-party studies on best practices for ethics, safety, fairness, inclusiveness, trust, and robustness for AI research and applications.
Another mission of the consortium will be to engage AI users and developers, as well as representatives from sectors such as healthcare, financial services, transportation, commerce, manufacturing, telecommunications, and media.
Oligopoly on AI?
The fact that some of the most invested companies in artificial intelligence research are partnering is probably an overall good thing. They will get to share research with each other, and advance AI that much faster, rather than operate in their own silos. They can also, potentially, act as checks and balances against each other before doing something dangerous with AI.
The biggest criticism this group is likely to face is the perception that what they’re actually trying to do is create an oligopoly (the market is shared by a limited number of competitors) on AI, not just a “partnership,” and ultimately try to kill or cripple emerging competition.
However, the group promises to open up the consortium to others and give them equal leadership roles. Unless they fail to live up to that promise, this is probably not something anyone needs to worry about too much.
What Privacy?
A second criticism would be related to privacy. If you were already worried about one of these companies having too much of your data, then them sharing all of their data with each other, ought to give you a panic attack. Google, Facebook, Amazon, and Microsoft’s services and products cover just about everything we do on our devices and online. The partnership already seems to have started with the wrong foot in terms of privacy, as the PartnershipOnAI.org site doesn't even use HTTPS encryption.
The companies haven’t mentioned any data sharing between each other yet, although that could be likely to happen in the future. AI is defined as much by smart algorithms as it is by data. The smarter the algorithms and the more data you have, the more advanced the AI becomes. Therefore, these companies sharing their users data with each other seems almost inevitable.
They may not do it automatically, but instead do what Facebook did with Whatsapp, and prompt users with a privacy policy update that says they can now share your data with the other companies in the consortium. The users’ choice is likely to be opt-out as well, as it was for WhatsApp users, because they know that opt-out choices means most users will agree by default without paying too much attention to the policy updates.
The companies promise to keep an eye on how they deal with privacy going forward as well. However, it's not clear what exactly that means, and whether they will even adopt privacy-friendly technologies such as "differential privacy" when mining users' data to improve their artificial intelligence technologies.
A Positive Perspective
This partnership could lead to many positive things, as well. Faster advancements in AI also means new cures for diseases, new materials and technologies that can bring interplanetary travel closer to reality, or simply significantly improved services.
If nothing else, OpenAI, the non-profit organization created by Elon Musk and a few others for the purpose of not leaving AI in the hands of a few powerful corporations or governments, seems to be quite excited about the Partnership on AI consortium, and hopes to be part of it soon, as well.
