Report text
Text extracted from the original PDF. Tables, footnotes, and reading order can differ. Check the PDF before quoting. Page numbers below refer to PDF viewer pages.
PDF page 1
Since Alan Turing posed the question “can machines think?” in 1950, engineers and technology enthusiasts have tinkered with artificial intelligence. AI is a class of technology mimicking human intelligence and problem solving. AI tools are now embedded into technology we use daily. Siri and other digital assistance use AI, as does predictive text in web browsers and email applications. Because AI is an emerging technology, there is a balancing act that must take place while considering regulations. Regulators must balance the ability for the private sector to innovate and develop AI with the threats AI could pose—disinformation, national security, identity theft, loss of control, discrimination and more. Better yet, regulators should trust the industries and experts using AI to implement best practices and standard of use before blanket regulations are introduced. Concerns and Consideration There is no one clear solution yet to regulate AI. Stakeholders are proposing frameworks for regulation, but as with emerging technologies and regulation, none have become the standout. This is mainly because of two conflicting interests–innovation and regulation. Too much regulation will stifle innovation; but too little regulation could fuel fears of an AI takeover. Moreover, there is a question of whether to regulate the AI technology itself, or to regulate the output. If regulators wish to create a sweeping framework to regulate AI, then regulators should keep the following considerations in mind. 1. Defining AI As an evolving and innovating technology in real time, the definition of AI will be imperative to regulators, developers, and users. Regulators will want it to encompass future technologies, so a regulation is in place when new tech comes online. Developers may advocate for a slightly more limited definition, so they are able to continue to create without immediate regulations. Users of AI will need a clear definition which is applicable across the board. AI does not stop at the state lines, so working to ensure there is a clear definition for all operating territories. The National Artificial Intelligence Initiative defines artificial intelligence as “a machine- based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments. Artificial intelligence systems use machine and human-based inputs to- (A) perceive real and virtual environments; (B) abstract such perceptions into models through analysis in an automated manner; and (C) use model inference to formulate options for information or action.” 15 U.S.C. §9401(3). 2. Transparency Since a reason for regulation is consumer protection and disinformation, an important piece of A Primer on Artificial Intelligence Regulations February 2024
Open PDF at page 1 ↗PDF page 2
AI regulations is transparency. Regulators want to ensure consumers have proper notice and knowledge about when they are interacting with AI tools, when AI manipulated an image, or AI wrote content. Another side to transparency is the data being uploaded and used by the AI tool. Users of AI tools must be able to trust the information coming from the AI, which comes from the data learned by the machine. So this data source must be accurate. Regulators may go so far as to say it must be known to the user what the data source is, but this could run into proprietary information both on the data side and the AI creation side. 3. Risk-based Proportionality AI tools are powerful and can be trained to complete complicated tasks. But, because AI can encompass a wide range of uses, regulators may consider regulating based on risk and reward. This is easier said than done, however, because defining or understanding what AI tools offer bad actors may yet to be discovered. Regulators would be attempting to regulate a problem that has not been created, begging the question, is AI ripe to regulate? 4. Intellectual Property Considerations Businesses creating and using generative AI are inputting information into the machine so the outputs will mirror its inputs. The inputs can include proprietary information and intellectual property. While proprietary information is typically protected as much as possible, intellectual property, especially copyrights, have constitutional right to protection. Lawsuits are being filed for the use of copyrighted material in generative AI’s input information, which in turn create AI outputs that contains protected information. To protect intellectual property rights, regulators must be prepared to strengthen copyright, trademark and patent laws to protect works from AI’s manipulation and use. In sum, AI regulation is still taking shape and is ever shifting based on new technologies. While there are current laws and regulations achieving the objectives of some regulators in the AI arena, they may need amending to ensure AI is included. For instance, consumer protection laws exist, expanding the current laws to ensure generative AI is included is much simpler than creating a new scheme, if the objective is to protect consumers from disinformation and deceptive trade practices. Regulators must consider if regulation of an evolving technology is ripe for a regulatory scheme that could stunt its innovation.
Open PDF at page 2 ↗