Abstract
The press and media, elected officials, some high-tech entrepreneurs, and a new breed of “ethicists” are calling loudly for governments to regulate artificial intelligence (“AI”), especially generative AI, popularized by ChatGPT and Google Gemini. Much of the clamor is ill-informed about the technology and based on exaggerated concerns about its risks. Nevertheless, the pressure for regulation is real and is already evoking governmental responses. Transparency requirements are far better than command-and-control regulation. The technology is far too new, however, for any form of command-and-control regulation. Would-be regulators need to take a deep breath and sit back for a while to see how the systems are used in the real world and what problems actually arise from their use, as opposed to rushing into action based on hypothetical nightmares.
Mandating transparency enjoys a long pedigree in regulation of technology. Arguments for greater transparency about the use of sophisticated AI technology are persuasive, but advocates of transparency should be more precise in exactly what they want to be transparent. Forcing developers and entrepreneurs to reveal the details of their black boxes—the statistical learning models and resulting algorithms—does little good to consumers and poses the greatest risk to legitimate intellectual property interests. Transparency with respect to the scope and content of machine learning (“ML”) databases, the fact of use, the purposes of use, the quality of results produced, and the availability of appeals to human beings is meritorious.