John Thickstun

Assistant Professor

John Thickstun, assistant professor of computer science at Cornell University, studies machine learning and generative models.

Thickstun says:

“This is an interesting perspective on safety that I might characterize as ‘safety as a service’ which allows organizations to outsource AI safety to a third party. These models are not quite a service for now (since they are released as open-weight models) but I imagine businesses like OpenAI might develop products offering safety and oversight functions for AI deployments in other organizations. I can imagine this being profitable for a safety provider and valuable to their customers as a kind of compliance/due-diligence process for deploying AI. I can also see issues with this safety paradigm, related to homogenization and centralization of safety standards across many industrial AI deployments by a single private organization.

“Safety is not a well-defined concept. Any implementation of safety standards will reflect the values and priorities of the organization that creates it, as well as the limits and deficiencies of its models. If industry as a whole adopts standards developed by OpenAI, we risk institutionalizing one particular perspective on safety and short-circuiting broader investigations into the safety needs for AI deployments across many sectors of society.”