Public trust in AI isn’t just eroding; it’s hardening into something more volatile. Recently, a 20-year-old allegedly attempted to firebomb OpenAI CEO Sam Altman’s San Francisco home. It was the most extreme incident yet but hardly an isolated one. In recent months, protesters have attempted to block the doors of OpenAI’s headquarters and staged hunger strikes outside Anthropic and Google DeepMind.
The numbers tell the same story: 66% of people globally already use AI regularly, but less than half actually trust this burgeoning technology we are beginning to rely on. Even as we’re encouraged to embrace medical diagnostics, autonomous vehicles and AI agents, industry leaders admit they aren’t sure how their own AI works. Even more surprising, as AI is already impacting jobs and careers, tone-deaf innovators are literally urging companies to “stop hiring humans.”
Rather than building dependable systems that work safely 99.999% percent of the time, many technologists are turning our world into a living lab without guardrails against catastrophic consequences. Every vendor makes loud commitments to responsible AI while overpromising their offerings, and the hype cycle is backfiring. Skepticism, confusion and fear only grow when technology fails to deliver, and when we cannot distinguish truth from PR and marketing buzz.
I fully embrace AI’s promise to improve our lives, and I believe we can pursue it responsibly and honestly. This is not one of those loud but empty commitments. At the University of Michigan, my team is designing VIGIL, an AI agent that will power mobile medical clinics and help address the rural healthcare crisis affecting 46.2 million Americans. It’s a crisis defined by aging populations, hospital closures and a shrinking medical workforce.
The Advanced Research Projects Agency for Health (ARPA-H) is addressing this challenge by investing in the development of mobile clinics. VIGIL is the AI system that will guide generalists, like family doctors and nurse practitioners, through specialist-level diagnoses and procedures so they can care for the patients before them.
With no room for error in medical care, our team is peeling off small problems one at a time to create AI systems that deliver with almost 100% percent accuracy. VIGIL empowers practitioners to interact directly with patients and make all care decisions.
Taking this a step further, VIGIL aims to strengthen the doctor-patient bond by employing the right technology, rather than the sexiest. For example, in place of VR goggles and laptops, which break eye contact, hide nonverbal cues and can lead to miscommunication and mistrust, our practitioners use large shared screens and a projector-based interface to literally project the screen onto the patient’s body and the examination table.
We’re building technology that earns trust by design, not by marketing.
Safety and trust should go hand-in-hand but the familiarity principle tells us that simple exposure to new technology turns fear into familiarity, and eventually, preference — and the AI industry is banking on this phenomenon.
When cars were first introduced in the late 1800s, it was a bit like 2022’s ChatGPT moment. Horseless carriages were exciting, yet they were seen as unreliable and potentially deadly. It took over two decades to build out an infrastructure of roads, traffic laws and insurance, and another decade for the technology to become mainstream.
After just a few years of whirlwind innovation, however, AI is already taking the wheel. Tesla and Waymo have yet to deliver the long-hoped-for L5 fully autonomous vehicle, but we’re already sharing roads with robotaxis and partially autonomous cars. AI’s warning signs are already here — the pitchforks in San Francisco are one of them.
We don’t let experimental medical treatments skip clinical trials because patients are desperate for a cure, and we shouldn’t let AI rush to market just because we are dazzled by its potential. As AI transforms our world, it’s time we prioritize real guarantees and transparency over hype and marketing gimmicks.
The car didn’t wait for the road, and people died until we built one. AI won’t wait either, but this time, we have the chance to build the road before the crashes start piling up. The public is already telling us that they don’t trust what we’re building. The industry can keep dismissing them, or it can start earning the trust it’s been demanding.
Dr. Jason Corso is co-founder and Chief Science Officer at Voxel51, a Toyota Professor of Artificial Intelligence, and Professor of Robotics and Electrical Engineering & Computer Science at the University of Michigan.
This article originally appeared on The Detroit News: AI industry has itself to blame for eroding public trust | Opinion
Reporting by Jason Corso / The Detroit News
USA TODAY Network via Reuters Connect
