An algorithm off switch isn’t enough. Big tech needs a duty of care over addictive designs | Zoe Daniel

Editor's take

The Guardian argues that merely providing users with an "off switch" for AI-driven features is insufficient to address the potential for addictive design. This perspective highlights a growing concern that AI, beyond its core functionality, can be engineered to exploit psychological vulnerabilities, mirroring issues seen in social media platforms. The implications extend to user well-being and the ethical responsibilities of companies like Meta, Google, and TikTok, who are increasingly integrating AI into their user-facing products.

The core issue is not just the presence of AI, but its deliberate application in ways that can foster compulsive engagement. This discussion is timely as AI models like GPT-4 and Gemini become more sophisticated and embedded in daily life, raising questions about how these powerful tools should be regulated. The call for a "duty of care" suggests a shift towards proactive ethical design rather than reactive user controls.

Moving forward, it will be critical to observe how regulatory bodies, such as the EU's Digital Services Act, begin to interpret and enforce such a duty of care in the context of AI. The development of concrete metrics or standards for evaluating "addictive design" in AI interfaces will be a key indicator of progress, as will the willingness of major AI developers to move beyond opt-out mechanisms towards fundamentally more user-centric and less manipulative design principles.