I Am Not a Robot

AI has been around for a long time—long enough that it has evolved into many different species, with different strengths and behaviors. I like to think of AI as a zoo of technologies. ... “AI” is just an umbrella term. Underneath it are dozens of different tools, systems, and species. And understanding those differences isn’t just nerdy taxonomy. It’s the key to knowing what these systems can actually do for you—and what they definitely can’t.

Margolies and other radiologists I spoke with said they wouldn’t want to work without the AI assistance now. The consensus is that these intelligent machines are helping human doctors perform better in three key ways. accuracy. ... confidence. The AI often confirms what a radiologist already thought, making the physicians more confident and relaxed. ... prioritization. In offices where there may be a backlog of mammograms or ultrasounds to read, the AI can prioritize the reports so that the more suspicious or possibly malignant findings are first in the line for radiologists to review.

An MIT graduate, Inam was living in Boston. She switched dentists after moving to a new neighborhood. The new provider gave her a treatment plan radically different from the one she’d received just six months earlier. “I asked for my X-ray, started reading up on Dental 101, and realized something was off,” she told me. “I gave my X-rays to different dentists and got different opinions—and I couldn’t tell who was right. It felt more like art than science.” That’s when she envisioned something like Overjet. “I said, hey, this data is unstructured. It’s so subjective. We can make the decision much more objective,” Inam said. “Then we can provide that information to patients so that they can understand their oral health and make the best decision for themselves.” And in many dental practices now, that is happening. But so is the opposite of Inam’s vision. These tools are also being used to justify more treatment—even when it might not be urgent or necessary.

Maybe the real story isn’t that AI will transform health care into something unrecognizable, but that it will act like an amplifier, exposing more information for medical practitioners and patients to evaluate and decide what is optimal or critical to address at the moment. This raises a harder question: Since humans built these health care systems in the first place, with all their human incentives and flaws baked in, shouldn’t we fix the systems themselves before unleashing machines that will only magnify the issues?

The sixty-year-old attorney, who now regularly rides in Waymos around Phoenix, rattled off the machine’s selling points like a closing argument: “They don’t speed. They don’t run red lights. They’re never going to be on a cell phone. They’re not going to be drunk. They don’t have a blind spot. They don’t get tired.” Lamber’s convinced the data will eventually show what he already believes: Autonomous vehicles are significantly safer than humans. Will there still be Waymo crashes that make headlines? Of course. But, he argues, imagine if every human-caused crash got that level of attention. “If this technology isn’t engaged in the human behaviors that cause accidents over and over again,” he said, “and because of that, lives are spared, injuries are prevented, society saves money on health care and insurance—why would we not get behind that?”

Home is where the hard is. And two things make the home particularly hard for robots—the unpredictability of the environment and the complexity of the tasks.

Moravec’s paradox, named for Hans Moravec, a pioneering roboticist at Carnegie Mellon University. In his 1988 book Mind Children, he wrote: “Unfortunately for human-like robots, computers are at their worst trying to do the things most natural to humans, such as seeing, hearing, manipulating objects, learning languages, and commonsense reasoning . . . Machines [do] well [at] things humans find hard, while doing poorly what is easy for us.”

Alpha Schools, a for-profit network of AI-powered private K–12 campuses in Austin, New York City, Northern Virginia, Tampa, and more. Students spend just two hours a day on core academics in the morning, delivered via AI tutors and adaptive learning platforms. The afternoons are dedicated to life skills, workshops, and project-based learning.

Teachers I spoke with said the best response is to shift what happens in the classroom. Instead of take-home assignments that can easily be outsourced to AI, they’re bringing deep-thinking and interactive work back into the class. That means having students write essays at their desks, holding more debates and discussions, and focusing on collaborative problem-solving.