Browse
Log in / Register
Share
It started as a pop song. In 2008, The Killers released "Human," and one line—"Are we human, or are we dancer?"—sparked grammar fights, internet memes, and endless commentary. On October 14, 2025, that question has outgrown the chorus. It now shows up in court filings, congressional hearings, and factory floors, as Americans try to figure out what artificial intelligence actually is—and whether it's becoming anything like us.
The lyric was never meant to be a test question. Brandon Flowers later explained he borrowed it from a hunter's remark about Americans' tendency to stare inward. Still, the phrase stuck because it touched something real: the line between who we are and what we perform. That blur is no longer metaphorical. Chatbots can mimic empathy, political leanings, and even romantic interest, which makes the song's question feel less poetic and more pressing.
The clearest sign the line is moving: how many people now treat chatbots as confidants. Apps like Replika and Character.AI have pulled in tens of millions of users, many of them teenagers and young adults. Teachers and parents have taken notice, and some families have sued chatbot companies over emotional harms allegedly tied to prolonged use. Psychologists who study technology have called this moment predictable. MIT professor Sherry Turkle warned in her book Alone Together that people would turn to machines for connection once real relationships got too demanding. People are now grieving chatbots that "die" in software updates and defending AI friends the way they would defend a human one. The attachment is genuine even if the relationship is one-sided.
Meanwhile, machines shaped like us have started doing human work. Figure's F.02 robot handles assembly tasks at a BMW plant in South Carolina, while Agility Robotics' Digit has been testing warehouse jobs arranged by Amazon. These humanoids walk, grip, and balance like people, but they never punch in, complain, or call in sick. That is exactly the pitch—yet it also forces a harder look at what work means. When a machine can sweat through the same shift, "labor" starts to sound less uniquely human. Automakers and logistics firms say the goal is to fill jobs people don't want. The quieter effect is on how we see ourselves: if the body is just a chassis, what does that make the worker inside it?
Law is where the question gets practical. In the long-running DABUS cases, an inventor named Stephen Thaler argued that an AI system should be listed as the inventor on patents. Courts in the United States and the United Kingdom said no: an "individual inventor" in the law means a natural person. The U.S. Copyright Office has also held that a work generated without human authorship can't be copyrighted. So legally, "human" is a clear status—and machines currently fail the test. But the rulings don't settle the debate. They only push it downstream. Europe has responded with the risk-based AI Act; the U.S. has no federal equivalent, only executive guidance, state-level bills, and voluntary industry pledges. Regulators are treating AI as a product, a risk, or a tool—never as a neighbor. That classification is doing a lot of hidden work.
Ask researchers what separates us from the machines, and you get less flattering answers: mortality, embodiment, vulnerability, and the ability to be wrong in genuinely original ways. Kate Darling, a research specialist at the MIT Media Lab, has argued that we project feelings onto robots far more readily than they return them. The danger isn't that machines feel; it's that we assume they do. An AI can model grief without grieving, can predict pain without suffering it. Philosophers add that humanness isn't a checklist of capabilities. It is a way of being that involves stakes. A machine doesn't fear its own end. It doesn't blush. It never has a stake in anything, because it never loses anything. Those qualities are exactly what make human choices matter.
For decades, the Turing test asked a deceptively simple question: can a machine imitate a human in conversation? That measure is now nearly useless, because imitation is the one thing machines do best. A more honest set of checks might look like this. First, consequences: would the system actually lose something—status, safety, love—if it made a terrible mistake? Second, originality: can it produce an error no dataset predicted, in a way that changes its behavior? Third, mourning: if the system were switched off tonight, would anyone genuinely grieve it, not merely miss the service it performed? By those tests, humans pass easily and current AI falls short. The point isn't to dismiss machines—it's to stop confusing intelligence with investment.
The stakes go far beyond tech blogs. This debate now shapes how children are raised around screens, how companies decide which workers to hire, and how courts decide who can be an inventor, an author, or a victim. Anxiety about our own distinctiveness runs underneath all of it. We worry that artificial intelligence will make us redundant, but we also worry it will make us merely predictable. The reassuring twist is that the question itself keeps failing to retire. That alone tells us something. We are the only species that asks whether it is the only one with a soul, a mind, or a reason to matter.
So, are we human? The honest answer, in October 2025, is that we are still deciding. The choice will be made in courtrooms and code, in classrooms and living rooms. What matters most is that people show up to the debate—and keep asking the question out loud. That habit, more than any benchmark, is what the machines can't copy.
