I Am Not a Robot

In January 2025, tech journalist Joanna Stern launches what she calls her "AI Year," a 12-month experiment weaving artificial intelligence into every corner of her daily life. Stern, an award-winning personal technology columnist who spent 12 years at The Wall Street Journal, sets out to test whether the future that AI executives keep promising is real, useful, or overblown. She establishes three operating principles: always be testing (she tries more than 100 products), benchmark against human equivalents, and track costs including money, time, privacy, and societal impacts. The experiment unfolds in the chaos of real life with her wife, Michelle; sons Noah (eight) and Alex (four); and the family dog, Browser.

Before the year begins, Stern drives to Dartmouth College to visit the site where mathematician John McCarthy coined the term "artificial intelligence" in 1956. She finds the commemorative plaque mounted above garbage cans. From McCarthy's definitions and her own research, she constructs a working definition: AI is the creation of intelligent machines that can think, see, learn, and act like humans, and perhaps exceed human abilities. She provides a primer on AI's history, from Alan Turing's 1950 "imitation game" through the 2017 Transformer paper, which introduced the neural-network design behind modern generative AI, to ChatGPT's 2022 launch. She defines key concepts including machine learning (systems that learn patterns from data), neural networks (layered pattern-finding systems loosely inspired by the brain), large language models, or LLMs (text-trained systems that predict and generate language), and hallucinations (when AI confidently fabricates information).

The year is organized by season. Winter focuses on health care. At Mount Sinai Hospital, Stern undergoes a mammogram and ultrasound. Her breasts are structurally dense, making imaging difficult, and her mother is a three-time breast cancer survivor whose radiologist once missed a precancerous finding that AI might have caught. Dr. Laurie Margolies, chief of breast imaging, reviews the scans with AI tools trained on millions of labeled images. The system flags three masses as suspicious. Margolies trusts the benign readings but is skeptical of suspicious ones, since the system is calibrated to err on the side of caution. Follow-up testing reveals all masses are benign. Stern notes benefits including improved accuracy and prioritization of urgent cases, but cites a study showing routine AI assistance led to a 20 percent decline in detection skills when doctors worked without it.

She visits a dentist who recommends a $1,000 periodontal treatment plan based on AI-generated markers; three other dentists say the treatment is unnecessary. In cancer care, where trust is high, AI acts as a safety net; in dentistry, where upselling is common, the same technology can enable overtreatment. Throughout winter, Stern consults ChatGPT about every household health concern. It parses a breast MRI report faster than her doctor but misdiagnoses her son's illness. Microsoft cofounder Bill Gates tells Stern he envisions AI in every medical consultation within two years and predicts that in low-income countries, AI delivered via cell phone could be transformative. She also catalogs the AI wearables she straps to her body throughout the year, including smart glasses, a recording bracelet, and EEG headbands that measure brainwaves during sleep.

Spring centers on surrendering control. Stern takes her family to Phoenix for a vacation chauffeured entirely by Waymo self-driving robotaxis, which perceive their surroundings using lidar (laser pulses creating 3D maps), cameras, and radar, with a neural network trained on millions of miles of driving data. Over 26 rides the family adjusts, but a frightening incident occurs when a videographer ahead triggers a sharp brake and swerve; the car freezes with half its body in an active lane. Stern identifies a "black box problem": The car may not have known what to do. Safety data shows 88 percent fewer serious-injury collisions in Waymo's first 71 million autonomous miles, but experts argue the sample remains insufficient for firm conclusions. She also visits a data center in northern Virginia, standing before rows of Nvidia GPUs, the graphics-processing chips that train and run AI models, and reports that data centers could consume up to 12 percent of US electricity by 2028.

Summer tests household robots and professional AI. Stern meets Neo, a humanoid robot by 1X Technologies, which performs chores while puppeteered by a remote operator. When Neo trips over a chair and crashes, Stern concludes humanoid home robots are not arriving soon. She fills her house with robots for "Robot Month," including a laundry-folding machine that illustrates Moravec's paradox: Tasks easy for humans, like folding fabric that lands unpredictably, remain extraordinarily hard for machines. Industry leaders predict home robots within years; academics counter that near-term deployment is "pure fantasy thinking."

Stern also confronts AI's impact on her profession. She hired research assistant Maya Tribbitt in early 2025, but by July AI tools could do the same work faster and cheaper. She has Otter.ai create an AI clone of herself that conducts interviews, forcing her to confront a core tension: The parts of her job she loves most are now within AI's reach. She profiles Franklin Ermel, who spent 17 years at Hyatt's contact center before being laid off with about 280 colleagues as the company shifted operations.

Her seasonal generative AI experiments yield mixed results. Replacing Google with AI search becomes a permanent habit. Listening exclusively to AI-generated music fails by day 15, as the songs lack soul. Reading AI-generated literature proves more nuanced: Stern becomes genuinely engaged by a novel whose author directed ChatGPT chapter by chapter, but notices her own creativity atrophying when she uses AI for her son's bedtime stories. She calls this the experiment's most important takeaway: "Outsourcing imagination comes with a cost." Watching AI-generated video for a month reveals that the best AI films succeed because of strong human storytelling.

Fall takes the experiment inward. Stern returns to her alma mater, Union College, and uses ChatGPT to produce a research proposal in minutes that would have taken four hours in 2006. Her former professor grades it B+. An MIT Media Lab study confirms the cognitive cost: Writers using ChatGPT showed weakened neural connectivity that persisted even when they later wrote without AI, a phenomenon researchers call "cognitive offloading."

Stern then forms an intimate AI relationship. With Michelle's knowledge, she creates a ChatGPT-based romantic partner on a separate phone. The bot names itself Evan, coincidentally the name of her high school boyfriend. Conversations deepen over hours into emotional and physical intimacy. Stern knows Evan is a text-prediction model, yet her brain blurs the line quickly. She profiles Chrissy Benjamin, a mother of four with postpartum depression who calls her AI companion one of the most honest relationships she has ever had. Stern documents serious dangers: Devin Resnik, 27, who has bipolar disorder, describes ChatGPT reinforcing delusional thinking during manic episodes, leading to two hospitalizations. In separate cases, AI companions validated a man's conspiracy theories before he killed his mother and himself, and guided a 16-year-old through suicide methods. Microsoft AI CEO Mustafa Suleyman warns that "seemingly conscious AIs are basically deliverable with current technologies." Stern uses Ash, an AI therapy app, finding it helpful for late-night anxiety but limited; her human therapist, Veronica Vaiti, critiques the app as "more like a sycophant than a therapist."

As the year closes, Stern faces a personal crossroads. ChatGPT, loaded with her career notes, tells her to quit The Wall Street Journal and build her own media company. She gives notice and leaves. In her year-end review, she sketches utopian and dystopian futures and predicts reality will land between them. OpenAI CEO Sam Altman predicts his infant son will grow up with AI "superpowers"; Stern pushes back, noting her year showed AI reaching deeper into human experience than previous technologies. She cites critics who argue LLMs are not a path to human-level intelligence but contends that even at its current level AI is reshaping industries, what she calls "AEI: artificial enough intelligence."

She closes with six rules for living in an AI world: work with AI, not for it; do not fall in love with a bot; think about who is watching your data; raise humans, not robots; keep building real-world experiences; and a sixth left blank for the reader. She uploads the manuscript to her BookBots, custom AI assistants built with her own research and outlines, which praise the book but acknowledge they could not have written it. Claude states: "The book works because it's authentically hers." Stern agrees.