SJ Bennett, one of the academics behind the six-person AI Ethics and Society team at the University of Edinburgh, perhaps best encapsulated the spirit of last Friday’s symposium on artificial intelligence, brokenness, and justice during the introductory talk.

“Today, we want to take brokenness as an entry point into the critical investigation of technology, flipping the dominant narrative of disruptive innovation and shifting the focus instead on what’s left broken in this process of disruption,” said Bennett to a hall composed of thinkers and researchers working at the intersection of contemporary technologies and morality. “Brokenness invites us to reflect on the social and technical mechanisms that operate selectively, and ask: These are disruptive for whom? Scholars and activists have long urged us to pay attention to how AI and other data-intensive systems can perpetuate marginalisation, displacement, and violence.”

While the various profound ways in which artificial intelligence—its formation, advancement, and usage by humans—negatively impacts people across the globe were discussed, what was also considered was how we collectively can go about care, repair, and resistance in reaction to this.

The day’s three investigative and collaborative panels began with Margins, Data Patchwork, and Justice, with Morgan Currie, Natassa Philimonos, and Srravya Chandhiramowuli. Though their particular interests and areas differ slightly, a large, overarching shared thread in their work is how contemporary technologies can obfuscate workers in inequitable ways. Chandhiramowuli, for instance, discussed how marginalised and often displaced communities in the Global South are doing the labour (often dubbed as “ghost work”) for AI to work—such as creating and annotating the datasets that are inherent to AI learning and expansion—and how deployed employment practices and rhetorics are further halting people in India, not least women, from gaining financial and personal independence.

Error, Uncertainty, and Categorisation, the symposium’s second panel, featured Alexander Campolo, Cindy Lin, and Benjamin Jacobsen. At the core of the discussion was the role of error, or outliers, in machine learning models: What can “errors” tell us about ourselves as a society? What does it mean when errors are perceived by humans as an aberration to be iteratively reduced and ultimately erased? “I think there’s an engineering idea which is: just get rid of errors,” Campolo explained—though, as he also later suggested, “‘Error’ is a concept that can illuminate what our cultures value as truth.” Relatedly, Lin mentioned how “Error helps us to create a context for understanding machine learning,” thereby helping us to contemplate questions such as “What do errors privilege as truth? […] What voices get seen and unseen?” Ultimately, machine learning “errors” themselves can highlight a wealth of insight regarding us as a society and the technology we’re developing, not least when it comes to entrenched biases, prejudices, and discriminatory views and actions.

Recommended reading

• UK Putting £8.5M Into Research That Helps Protect Society From AI Risks
• Scots Unis to Take Part in £3.5M Project to Reduce AI Harm
• Navigating the Ethical Implications of AI ‘Deadbots’

The final panel, Care, Repair, and Craft, included Anne Lee Steele, Alex Taylor, and SJ Bennett. Considering artificial intelligence’s brokenness, how can repair happen? How can we better care for one another, and go about the caring, amid this fragmentation? During this collaborative discussion, the themes—and modes of response—that the panellists kept returning to were bulding human connection and community organising. As a way of putting the public’s voice to the fore, Taylor touched on his current work with BRAID—the Bridging Responsible AI Divides programme, led by the University of Edinburgh in partnership with the Ada Lovelace Institute and the BBC. As part of a BRAID research project, Taylor and a team are doing on-the-ground work in multicultural Leith, hearing and understanding the first-hand thoughts and opinions of local residents. As Taylor explained, “It’s asking the people of Leith: What is AI for you? How can it make a difference to you?”

The day’s collection of dialogues and questions served as something of a much-needed repositioning of how we may consider our current predicament with artificial intelligence. While, recently, it’s felt like humans’ development of artificial intelligence is something akin to a speeding train without working brakes—a turbulent force that cannot and will not be slowed down—AI, at least not yet, isn’t a fully autonomous entity. Human beings still remain behind AI and its implementation on all levels—how it’s created, used, and advanced. Within this there is possibility: As those using, interacting, and living with AI, we can try to use our collective voices, efforts, and actions to not only care for others impacted by its negative consequences, but to also reach the people behind AI’s development and help repair the broken machine.