US tech policy needs to keep pace with AI innovation

Image credit: Ole_CNX / Getty Images

As artificial intelligence (AI) innovations outpace the news cycle and grab the attention of society, this unprecedented wave of technology will reach its full potential as a positive contribution to economic and social progress. To that end, frameworks for the responsible and ethical development and use of AI are becoming increasingly important. .

The European Union is already working on legislation on responsible AI. I shared my thoughts on these efforts nearly two years before him. And the AI ​​Act, as it was known, was “an objective and prudent approach to innovation and social considerations.” Today, technology company leaders and the U.S. government are coming together to develop a unified vision for responsible AI.

The power of generative AI

Last year, OpenAI launched ChatGPT, captivating the imagination of technology innovators, business leaders, and the general public, sparking an explosion of consumer interest and understanding of the capabilities of generative AI. However, the rise of artificial intelligence into the mainstream, such as in political issues, and the human tendency to experiment and test systems, quickly raises the potential for misinformation, privacy implications, and risks to cybersecurity and fraud. You run the risk of falling behind.

In an early effort to address these potential challenges and ensure responsible AI innovation protects the rights and safety of Americans, the White House has announced new measures to advance responsible AI. bottom.

In a fact sheet released by the White House last week, the Biden-Harris administration outlined three actions to “promote America’s responsible innovation in artificial intelligence (AI) and protect people’s rights and security.” These include:

• New investments to advance responsible American AI research and development.
• Public evaluation of existing generative AI systems.
• Policies to ensure that the U.S. government leads by example in mitigating AI risks and capitalizing on AI opportunities.

New investment

In terms of new investments, the National Science Foundation’s $140 million funding to launch seven new national AI labs pales in comparison to funding raised by private companies.

While the direction is right, the U.S. government’s investment in AI has generally been paltry compared to other government investments, namely China, which began investing in 2017. A pressing opportunity exists to extend the impact of investments through academic partnerships for workforce development and research. Governments should fund AI centers alongside academic and corporate institutions that are already at the forefront of AI R&D, driving innovation and creating new opportunities for businesses with the power of AI. is.

Collaborations between the AI ​​Center and top academic institutions such as MIT’s Schwartzman University and Northeastern’s Experimental AI Lab bring together experts from academia, industry, and government to collaborate on cutting-edge research and research. and help bridge the gap between theory and practice. A development project containing a real-world application. By partnering with leading companies, these centers can help businesses successfully integrate AI into their operations to improve efficiency, reduce costs, and improve consumer outcomes.

Additionally, these centers will help educate the next generation of AI professionals by giving students access to cutting-edge technologies, hands-on experience in real-world projects, and mentorship from industry leaders. By taking a proactive and collaborative approach to AI, the U.S. government can help shape a future in which AI enhances, rather than replaces, human work. As a result, all members of society can benefit from the opportunities created by this powerful technology.

public evaluation

Model evaluation is important to ensure that AI models are accurate, reliable, and unbiased, and is essential for successful deployment in real-world applications. For example, imagine an urban planning use case that trains generative AI in red-light cities with historically underestimated poverty populations. Unfortunately it just causes more of the same. The same goes for biases in lending, as more financial institutions use AI algorithms to make lending decisions.

If these algorithms are trained on data that discriminates against certain demographic groups, they can unfairly deny loans to those groups, leading to economic and social disparities. These are just a few examples of biases in AI that should always be kept in mind, regardless of how quickly new AI technologies and techniques are developed and implemented.

To combat AI bias, the government will create new opportunities for model evaluation at DEFCON 31 AI Village, a forum for researchers, practitioners and enthusiasts to gather to explore the latest advances in artificial intelligence and machine learning. Announced. Model Evaluation is a joint initiative with leading companies in the field such as Anthropic, Google, Hugging Face, Microsoft, Nvidia, OpenAI, Stability AI, and leverages the platform provided by Scale AI.

In addition, it measures how well the model fits into the principles and practices outlined in the Biden-Harris administration’s AI Bill of Rights Blueprint and the National Institute of Standards and Technology (NIST) AI Risk Management Framework. . This is a positive development as the administration engages directly with companies and leverages the expertise of a technology leader in the field who has become his AI lab at the companies.

government policy

Regarding the third action on policies to ensure that the U.S. government is leading by example in mitigating AI risks and exploiting AI opportunities, the Office of Management and Budget issued a public comment on the use of AI systems by the U.S. government for public comment. We plan to draft policy guidance. . Again, no timeline or details for these policies have been given, but it is expected that the Executive Order on Racial Equality issued earlier this year will be at the forefront.

The executive order includes provisions directing agencies to use AI and automated systems in ways that promote fairness. For these policies to have a meaningful impact, incentives and spillovers must be included. It should not be just arbitrary guidance. For example, his NIST standards for security are valid requirements for deployment by most government agencies. Failure to comply with these is highly embarrassing, at least to those involved, and is grounds for personnel action on the part of the government. To be effective, government AI policies as part of NIST and others must be comparable.

Moreover, the cost of complying with such regulations should not be a barrier to startup-led innovation. For example, what can be achieved in a framework in which the cost of regulatory compliance varies with the size of the business? Finally, as governments become more important buyers of AI platforms and tools, their policy Most importantly, it serves as a guide for building Making adherence to this guidance literal or effective purchase requirements (such as FedRamp security standards) could make a big difference to these policies.

As generative AI systems become more powerful and pervasive, all stakeholders, including founders, operators, investors, technologists, consumers, and regulators, will deliberately and deliberately pursue and engage with these technologies. it is essential to Generative AI and broader AI have the potential to revolutionize industries and create new opportunities, but they also pose significant challenges, especially around issues of bias, privacy and ethical considerations.

All stakeholders must therefore prioritize transparency, accountability and collaboration to ensure that AI is developed and used responsibly and beneficially. This means investing in ethical AI research and development, engaging with diverse perspectives and communities, and establishing clear guidelines and regulations for the development and deployment of these technologies.