According to a July report from the ed-tech publication Tech & Learning, at least 15 U.S. states have introduced bills to study how AI is used across government. Among them is Connecticut’s SB 1103, which would require the state’s Department of Administrative Services to assess AI programs and systems being used by state agencies, as well as a resolution in Louisiana asking the state’s Joint Legislative Committee on Technology and Cybersecurity to study the impact of AI tools in the public sector. At the federal level, the Department of Education released a report recently that urged pragmatism when it comes to weighing the risks and benefits of AI ed-tech tools and recommended that federal policymakers work more closely with states, higher-ed institutions and K-12 schools to develop and assess guidelines specific to the use of AI in education. The report also recommended that schools and universities involve educators in discussions about tech procurement.
Consortium for School Networking (CoSN) CEO Keith Krueger wrote in an email to Government Technology that discussions about regulating AI in education are still largely in their infancy at the state and federal levels. However, he said, state and federal lawmakers may play a critical role in giving schools and universities direction on how to make use of AI.
“Clearly federal standards will be helpful for national consistency, but may take a while to happen,” he wrote. “Therefore, it makes sense for states/districts to understand and require disclosure by companies in how they are using AI. We welcome states’ desire to better understand the landscape — as demonstrated by common focus on studies — before rushing to regulate the technology’s use by educators.”
Krueger pointed to the White House’s Blueprint for an AI Bill of Rights — which outlines some of the risks and benefits of using AI in fields like law enforcement, health care and education — as a good foundation for how to think about the technology amid today’s policy discussions.
“Guardrails around ethical use of generative AI in education, commerce, health industry and government will likely call for federal regulations, licensing, and the creation of an independent commission to study the AI environment and consider the creation of a U.S. AI agency aligning to global efforts,” he wrote. “As we know, the wheels of government do not move fast, yet AI is moving at warp speed.”
As we know, the wheels of government do not move fast, yet AI is moving at warp speed.
CoSN CEO Keith Krueger
“It’s encouraging to see New Jersey, Connecticut and Louisiana acknowledging the profound potential of this technology and attempting to proactively explore AI’s impact on their state agencies’ operations, procurement and workforce readiness,” Fallon wrote. “As we embrace AI-powered tools in education, it’s crucial that all students, regardless of their background or location, have equitable access to these resources. State education agencies can work to bridge the digital divide by allocating resources to underserved communities, ensuring that AI technologies reach every corner of the education system.”
Fallon said one of the first major steps in formulating policies at the institutional, district, state and federal levels about how to use AI tools safely and ethically is understanding how the technology works, as well as collaboration between educators, policymakers, technology developers and other stakeholders to make sure particular AI tools align with the specific needs of the school district. She said it’s also important to take digital privacy and cybersecurity concerns into consideration when adopting and deploying AI tools.
“Data privacy and security are non-negotiable. States should establish stringent guidelines for the collection, storage and use of student data within AI systems,” she wrote. “Protecting students’ privacy is paramount, and policies must be in place to safeguard their sensitive information while still enabling meaningful use of AI-driven insights.”
Neil Heffernan, a computer science professor at Worcester Polytechnic Institute and developer of the AI-driven homework assistance tool ASSISTments, said future AI regulations should avoid stifling research and development efforts. He added that he believes recent calls from tech companies such as ChatGPT-developer OpenAI to regulate the technology could lead to the monopolization of AI research and development, which could ultimately impede tech advances.
“OpenAI knows one way to cripple the open-source movement of using large language models — like LLaMA 2 that Meta put out last week — is to enforce all sorts of [rules that place] regulatory burden on the little guys. OpenAI would love to say you cannot use LLM [tech] unless you have done a $100,000 fairness test for bias. OpenAI can so easily pay that fee but the open-source folks can’t,” he said.
John Bailey, a senior fellow with the policy think tank American Enterprise Institute, echoed Heffernan’s concerns. He added that since AI technology is still fairly new, it’s difficult to say just how AI tools should be regulated.
“I’d encourage a bit of a cautionary approach about regulating this. I think we’re so early into the [development of] the technology, and it’s changing at such a rapid pace,” he said. “With that kind of speed, it’s easy to get the regulation wrong in many respects.”
