At Data Summit Connect 2021, David Weinberger, author of Everyday Chaos: Technology, Complexity, and How We're Thriving in a New World of Possibility, reflected on how machine learning is changing our ideas not only about how knowledge is developed and deployed, but even about what knowledge is and who we humans as knowers are.
Increasingly, machine learning models will be our houses of knowledge, the way libraries and human minds are now. Yet there is no knowledge inside a machine learning model, says Weinberger, who notes that this will mean rethinking the nature and use of knowledge, and even who we are as the creatures that can know their world.
In his wide-ranging talk, Weinberger explored what it means to get knowledge from a system that actually has no knowledge in it, and whether there is the capacity for there to be a combination of human values and machine skill to avoid black-box situations and keep a human in the loop. He concluded with a suggestion for a how there may be some interoperation between our old systems of knowledge, which incorporate nuance and perspective, and the new systems of AI.
Weinberger's proposal for a strategy that enables overlap between our age-old concept of knowledge and newer approaches provided by algorithms, AI, and the answers they provide is "the centaur."
For example, he said, he has been asked, frequently especially among librarians, whether machines will replace human workers.
Similar to the concept of a chess centaur which merges the human experience and instincts of a grand master with a computer’s unfailing ability to calculate strategies and outcomes, "a social centaur" could enable cooperation between humans with empathy and interest and the computer’s precise knowledge of authors and subject matter.
Unlike an online book seller whose only interest is making a sale, no matter what that may be, in the example of the library setting, Weinberger said, there is concern for addressing constituents’ needs in a more profound and personal way. As a result, he suggested, there is room for the social centaur, in which librarians and AI work together with a librarian using AI recommendation systems to address the importance of culture, politics, and the need for people to not just have their requests met, but to also expand their experience beyond what is comfortable, familiar, and expected.
This sort of centaur works best and is most plausible, Weinberger said, when the differences between the machine and the people have to do with an intersection of values and culture provided by humans and the ability to handle scale and to provide a set of skills and capabilities—provided by machine learning.
Weinberger’s session was titled “Knowledge in the Age of AI.”
More information about Data Summit Connect 2021 is available here.
Replays of this and all Data Summit Connect 2021 sessions are available to registered attendees for a limited time.