EXPERT INSIGHT: Congress Explores ChatGPT and Artificial Intelligence. Will America First Solutions Prevail?
Artificial intelligence, including ChatGPT, is cutting-edge technology that has the potential to improve numerous aspects of modern life.
However, these tools have the potential to be weaponized against the American people if left unchecked. Meanwhile, overregulation could stifle innovation and put the U.S. at a competitive disadvantage with its global adversaries.
Congress and private industry need to establish America First regulations of AI to enable society to reap the benefits of this powerful technology, but they should also install safeguards that prevent it from harming American institutions.
Throughout our Nation’s history, the U.S. has been a leader in developing cutting-edge technology, processes, and products that improved the lives of millions. Countless examples include Henry Ford’s streamlined process for building cars, the Internet, and Steve Jobs unveiling Apple’s iPhone. Today, the latest technological advancement creating a global buzz is generative artificial intelligence (AI), which includes OpenAI’s ChatGPT, which was released in November 2022.
Many Americans are now asking: What is ChatGPT? One detailed definition is the following: “ChatGPT is an advanced conversational AI language model developed by OpenAI. Based on the GPT-3.5 architecture, it utilizes deep learning techniques to generate human-like responses in natural language conversations. The model’s ability to understand and generate coherent and contextually relevant responses has significantly advanced the field of conversational AI, enabling a wide range of practical applications.”
The sentences above were generated by ChatGPT when prompted to “Describe ChatGPT for a short research paper.” If you are surprised to learn this, you are likely not alone. The response, generated in a matter of seconds, included a title for the paper, “ChatGPT: An Advanced Language Model for Conversational AI,” as well as eight suggested sections of the paper with brief descriptions of each. Part of the summary of the conclusion section reads, “ChatGPT has emerged as a powerful conversational AI model, showcasing the potential of deep learning and natural language processing in generating human-like dialogue…. With ongoing research and improvements, ChatGPT is expected to play a pivotal role in revolutionizing the way humans interact with AI systems.”
As this simple example shows, ChatGPT, including its latest update known as GPT-4, can inject efficiencies into research and other queries. However, in its current unregulated form, numerous downsides have quickly emerged. Students have begun relying on ChatGPT to write papers and assist with homework and exams, creating a massive cheating scandal in school districts across the country. Though ChatGPT and generative AI are highly sophisticated, they are imperfect and have occasionally churned out misinformation. Additionally, another implication is that the efficiencies afforded by ChatGPT and similar AI will make certain jobs obsolete, with one estimate predicting that 300 million jobs worldwide could be negatively impacted by this new technology. On the other hand, new jobs will emerge as this technology continues to develop and improve, but it is unclear how many Americans who are displaced by AI develop the skills necessary to obtain new employment in this emerging industry.
Against this backdrop, on May 16, 2023, the Senate Judiciary Subcommittee on Privacy, Technology, and the Law held a hearing titled “Oversight of A.I.: Rules for Artificial Intelligence.” The witnesses testifying at the hearing included Samuel Altman, CEO of OpenAI; Christina Montgomery, Chief Privacy & Trust Officer of IBM; and Gary Marcus, Professor Emeritus at New York University.
Unlike many congressional hearings that devolve into grandstanding to make political points to be featured on cable news shows, this subcommittee hearing was a bipartisan inquiry focused on a better understanding of ChatGPT and AI and how Congress should deal with this revolutionary technology. Altman testified that “AI has the potential to improve every aspect of our lives” but conceded that it is “unusual tech” and that safeguards are needed. Meanwhile, Professor Marcus, an AI critic, raised numerous concerns about misuse of the technology, including micro-targeted ads for pharmaceuticals and election misinformation. The hearing also explored questions regarding intellectual property, as Senators Marsha Blackburn (R-TN) and Mazie Hirono (D-HI) expressed the need to protect the intellectual property rights of musicians and other artists from being misappropriated by others who may take credit or profit for “original” songs or designs generated by AI that copy preexisting work products.
Although there appeared to be a general consensus at the hearing that some degree of government regulation is appropriate, views varied on what that looks like. In simple terms, the regulatory options run the spectrum from completely unregulated to bureautic micromanagement. However, as real-world examples have already demonstrated, a hands-off, laissez-faire approach has too many negative drawbacks. For example, on May 22, an AI-generated image that purported to show the Pentagon on fire circulated on Twitter through a blue check “verified” account. As this fake news spread, the S&P 500 dropped .3%, demonstrating that unregulated AI has the potential to move markets and wipe out value.
Additionally, impersonating world leaders, politicians, and business executives is another legitimate concern regarding unchecked AI. Chairman Senator Richard Blumenthal (D-CT) revealed that his opening remarks were drafted by ChatGPT and “spoken” by a computer-generated voice that sounded like the senator to illustrate this point. The statement began by saying, “Too often, we have seen what happens when technology outpaces regulation. The unbridled exploitation of personal data, the proliferation of disinformation, and the deepening of societal inequalities.”
At the other end of the spectrum, over-regulation will stifle innovation in America and enable adversaries like China to gain a competitive global advantage by using AI. If the federal government bans AI completely, American businesses will lose out on the research efficiencies a product like ChatGPT offers, the opportunity to launch new industries, and the wealth generation that comes with them. The answer appears to the question regarding regulation lies somewhere between these two ends of the spectrum. The witnesses at the hearing offered the following suggestions for regulatory intervention:
- Sam Altman: require a government license for AI products above an undefined scale of power.
- Professor Marcus: create a government entity similar to the Food and Drug Administration that is charged with pre-reviewing and approving AI products before they are introduced to the market, and establish a separate new government entity that is “nimble” and able to track the AI product for abuse post-release.
- Christina Montgomery: implement an approach similar to the European Union’s European AI Act that categorizes AI applications into four levels of risk (unacceptable risk, high risk, limited risk, minimal or no risk) with safety obligations proportional to the degree of risk. Montgomery noted that IBM supports this proposal.
The hearing on May 16 was more of an informational probe, but the committee’s attention to this matter signaled that it was focusing on the right issues at the right time. The realm of AI is developing quickly, and Ranking Member Josh Hawley (R-MO) noted that the hearing could not have occurred at this time a year ago because the technology did not exist. It is premature to know whether any of the “middle ground” regulatory options offered by the witnesses are appropriately tailored based on a market cost-benefit analysis. A thorough analysis would enable policymakers to determine what they can do to protect the American people from this technology, enable businesses to use it effectively and enable our Nation to benefit most financially in a responsible manner. What is abundantly clear is that lawmakers currently lack sufficient understanding of AI, including ChatGPT, and that further research is needed to determine what an America First solution to ChatGPT and AI looks like. However, this congressional inquiry was a positive step in the right direction that will set the stage for more discussions on this critical subject.