In our age of hyper-partisan politics, one area that seems to be attracting notable bipartisan congressional concern, including various potential legislative approaches, is the real-time development and implementation of artificial intelligence. AI’s reach across many economic sectors and its effect on education, medical research and national security poses complex legal, social and moral questions that need to be addressed.
The Senate’s hearings in May clearly demonstrated that Democrats and Republicans were eager to learn more. As Sen. Josh Hawley, R-Missouri, noted, “We could be looking at one of the most significant technological innovations in human history.” At the other end of the political spectrum, Sen. Dick Durbin, D-Illinois, clearly agreed with Hawley’s assessment regarding what is at stake. “The magnitude of the challenge … is substantial. I’m not sure that we respond quickly and with enough expertise to deal with it.”
There are no legislative solutions to some of AI’s significant potential problems. According to Hawley, these include loss of jobs, invasion of privacy on a scale we have never seen, manipulation of personal behavior, manipulation of personal opinions, and potentially the degradation of free elections.
Consequently, an important AI-related item — the re-establishment of the Congressional Office of Technology Assessment — needs to be tucked into the federal budget authorization to be enacted before the end of September for government funding to be available for the new fiscal year.
The OTA was extinguished 28 years ago, based on the premise that Congress did not need better expertise when evaluating the technological implications of legislation in its pipeline. The Office of Technology Assessment received $22 million annually at its peak funding. Its defunding was done with a mistaken sense that even this minuscule amount was too much to spend to ensure that Congress could remain up to speed on technological developments. The coming AI era has demonstrated clearly the unmindful nature of this decision.
With an OTA, Congress would have been more knowledgeable about all aspects of artificial intelligence, particularly fast-paced developments in large language machines that process data for use in AI systems and generative AI platforms such as ChatGPT. Hand-wringing about the decision to shut it down, however, only looks backward in speculating about previous decisions regarding how to spend taxpayer dollars. Congress should now rectify its “penny-wise and pound-foolish” approach by ensuring the OTA is re-established as soon as possible.
Any funding allocation still would be tiny, not even at the level of any federal budget rounding errors, yet the payoff would be substantial. How will Congress effectively allocate such massive amounts for AI guidance that requires detailed information on the technologies that comprise its elements? How will Congress provide meaningful oversight for these funds once they are authorized? These questions must be addressed if a proposal to re-establish the OTA is forthcoming.
Indeed, the Select Committee on the Modernization of Congress in 2019 favorably reviewed whether a renewed Office of Technology Assessment would help provide better expertise to lawmakers. But only $6 million was included in a legislative spending bill to do this, and even this amount was not forthcoming. As the Belfer Center for Science and International Affairs at Harvard University correctly noted,“Congress does not, in fact, lack a supply of folks trying to give it advice. To the contrary, it is overwhelmed. What it lacks is the ability to sort through it all.”
From this perspective, the Office of Technology Assessment should have a reconstituted mission to recognize the value of new, further external information about technology, assimilate it, and assist Congress with applying it to legislative outcomes.
The Office of Technology Assessment can deliver the best bang for the buck if it can help our lawmakers with the avalanche of information about AI generated by universities, think tanks, nonprofit groups and the private sector. Some might argue that other government agencies, such as the Congressional Research Service and Government Accountability Office, can assume this role if provided additional funding and staff, therefore minimizing the possibility of including duplicative resources.
These impressive organizations have a broad range of topics to cover and are designed to generate new research rather than synthesize the best available technology knowledge for Congress. A lean and mean Office of Technology Assessment, funded for an initial period of five years with a sunset provision to make it politically palatable, would be a far more attractive option. Whatever the price tag, it holds the promise of being the greatest bargain for our nation than any other prerequisite to AI legislation could offer.