4 minSociety
AI's Potential to Help, Not Just Harm, Draws New Attention
From restoring speech for a man with ALS to detecting breast cancer earlier, a growing body of examples shows that artificial intelligence can be designed to serve people — if humans make that choice.
Artificial intelligence is often framed as a looming threat, but a growing set of real-world examples suggests the same technology that raises fears about job losses and misinformation can also restore abilities, improve medical diagnoses and make digital tools accessible to far more people. The difference, according to designers and researchers, lies in the choices made by the people who build it.
One of the most striking cases involves Casey Harrell, a climate activist whose speech had been slowly taken by ALS, a progressive neurological disease. By the time even those closest to him struggled to understand him, surgeons at the University of California, Davis implanted tiny sensor arrays in the part of his brain that controls speech. AI learned to interpret the signals his brain sends when he tries to talk, converted them into words and spoke them aloud in a voice rebuilt from old recordings. The first time he heard it, he said it felt a lot like himself. People who had not heard his voice in a long time cried.
The medical field offers other examples. In Sweden, Lund University ran the MASAI trial, in which roughly 106,000 women in the national screening program were split into two groups: half had their mammograms read the traditional way, and half were read with AI support. The AI-assisted group saw 29 percent more cancers detected, and final results showed 12 percent fewer cancers slipping through between screenings, including fewer of the aggressive kind.
Beyond cancer screening, AI has accelerated work in biology. For 50 years, determining the three-dimensional shape of a single protein could take a scientist years. In 2020, DeepMind's AlphaFold learned to predict those shapes from a sequence alone. Two years later, its team released predicted structures for more than 200 million proteins, and the tool has since been used by over three million researchers in more than 190 countries. Its creators won the Nobel Prize. At MIT, researchers trained an AI to scan more than 100 million chemical compounds for ones that could kill bacteria, finding halicin, which killed strains resistant to every known antibiotic in lab tests, and later abaucin.
Accessibility is another area where the technology is beginning to flip the usual dynamic. For most of computing history, people have had to adapt to the machine. Be My Eyes, founded in Denmark by a blind furniture craftsman, began as an app connecting blind people with sighted volunteers over video. In 2023 it added Be My AI, which lets a user point a phone at a menu, a medicine label or an outfit and receive a detailed description back. Harrell's case represents the far end of the same idea: an interface that adapts so completely it meets a person inside their own body.
Artists and creators are also finding new uses. A friend of one designer described how low-income filmmakers in his community can now tell their stories thanks to accessible generative AI tools for film, potentially opening access to funding that had been limited to privileged groups. The designer noted that AI now makes it easier to reformat writing, convert handwriting into digital documents and automate the organizational layer of a creative process, freeing time for actual creation.
The common thread, according to this view, is that technology is never neutral. Every product is a stack of decisions about who it serves, what it rewards and what it refuses to do. A feed that keeps users scrolling at 2 a.m. is a design choice, just as a tool that hands someone back their evening is. The good outcomes are not lucky accidents. In each case, someone decided who the tool was for, put the right people in the room and kept humans in charge. The question, proponents argue, is whether more of those choices will be made.
4
