In an effort to regulate use of emerging technology, the General Services Administration recommended that agencies looking into buying generative AI services and related hardware should weigh its benefits. Alex Major, Co-Chair of McCarter’s Government Contracts & Global Trade Practice, spoke with Law360 and said that agencies’ potential interest in using “a bright and shiny toy” might not be appropriately tempered by considerations of risk when determining whether to use generative AI. “It’s really an attractive nuisance in a lot of ways,” Alex said. “And because of that, the guardrails, just like with any attractive nuisance … need to be very, very clearly identified for the government. These tools are so incredibly dynamic, not taking the time to put in [specific] guardrails is, I think, beyond risky.” There are some more descriptive aspects of the guidance, however, that can meaningfully help agencies in determining whether using generative AI makes sense for their needs, such as the GSA’s suggestion that agencies use a “sandbox” — a type of small-scale, low-cost technology test bed — before committing to any full-scale purchase of an AI system, Alex added.
5.8.2024