Dean Ball serves as the State and Local Policy Senior Program Manager, the Hoover Institution, and as board member, The Alexander Hamilton Institute for the Study of Western Civilization (AHI). Mr. Ball received a grant to write a column on artificial intelligence (AI). “Hyperdimensional,” an online newsletter, is a result of this effort.
Mr. Ball provides an ongoing analysis of issues in emerging technology. His most recent piece, “The Case for Public AI Infrastructure,” makes the point for an important—but limited—federal role in AI. Governmental policy has focused too much on “conjectures” about its potential dangers, he observes, rather than steps to prepare society for the profound and unstoppable fallout of AI.
“[T]echno-optimists and classical liberals … often spend their time … worrying about the unintended consequences of regulating [AI] too much and too early. This work is critical, no doubt, but it can leave one with the impression that there is nothing to be done other than to let the market rip.” It would be better, Ball suggests, for the federal government to focus in other areas by helping, for example, to create “a publicly accessible AI research infrastructure.”
Among the questions that federal resources for such infrastructure can address: “AI alignment (how consistently a model obeys human values … ), steerability (user control), and interpretability (understanding of how a model works),” all of which are “large unsolved scientific problems.” There must, Ball says, be “as many minds … as possible” dealing with them, and academic institutions “lack the computational resources to allow [their] people to do their best work” in this area.
Leave A Comment