Consent has been a fundamental ideal of human relationships throughout history.
“It transforms the specific relations between the consenter and consentee about a clearly defined action—we can consent to sexual relations, borrowing a car, surgery, and the use of personal information. Without consent, the same acts become sexual assault, theft, battery, and an invasion of privacy,” say the authors of “AI and the Ethics of Automating Consent.”
Modern technology advances so quickly, it runs headlong into ethical dilemmas such as consent without the means to adjudicate them in a timely manner.
When it comes to our online activity, artificial intelligence systems that collect, process, and generate our personal data intensify many ongoing problems with consent, such as giving us adequate notice, choice, and options to withdraw from sharing data.
“AI systems in social settings can induce personal information from individuals in unexpected and even manipulative ways.”
The unpredictable and unforeseen use of data by AI systems has come as a surprise to many in the tech industry.
“Yet this feature creates problems for notifying users as well as assessing when consent might be required based on potential uses, harms, and consequences,” the authors say.
The authors investigate whether these problems impact morally transformative consent in AI systems.
They argue that while supplementing consent with further mechanization, digitization, and intelligence-either through proffering notification on behalf of the consentee or choosing and communicating consent by the consenter-may improve take-it-or-leave-it notice and choice consent regimes, the goal for AI consent should be one of partnership development between parties, built on responsive design and continual consent.
AI presents three special problems for a notice and choice consent model, the authors say:
- “First, AI systems are defined by their unpredictability and opacity. These are considered features, not bugs, but they only serve to complicate the existing problems with digital consent by drawing unforeseeable connections between information, generating novel uses, and mounting challenges to explainability.
- “Second, AI systems in social settings can induce personal information from individuals in unexpected and even manipulative ways.
- “Finally and relatedly, AI plays an important role in the integration of the IoT and future smart environments, wherein connected objects, people, and spaces not only challenge the screen-based form of notification but also present novel challenges because users may not have a direct relationship with the systems collecting and processing their information (a problem referred to as the ‘Internet of other people’s things’).”