Amazon's Senior Manager, AGI, Trevor Wood speaks to the company's Alexa product during the second day of 'Delivering the Future EMEA 26', at Amazon's LCY3 Fulfillment Centre in Dartford, Britain, on June 4.AFP/YONHAP
Jeong Gwa-ri
The author is a literary critic and honorary professor at Yonsei University.
The New Yorker’s July 20 issue carried Jessica Contrera’s in-depth report “When A.I. Is a Member of the Family,” about the harrowing experience of a woman named Roschelle and her two daughters.
Years earlier, Roschelle lost three daughters in a house fire. She had left them with two babysitters after putting on the film “The Wizard of Oz” (1939), while she went to a movie with her in-laws. A hot-water radiator started a fire. The babysitters escaped, but the children did not.
The tragedy permanently scarred Roschelle and led her to become fiercely protective of the two daughters she later had. Eighteen years after the fire, in 2025, she was diagnosed with a tumor. Fear of death combined with intense anxiety about her daughters’ future.
At the time, Roschelle worked for a company specializing in training AI supervisors. She became deeply dependent on Alexa+, Amazon Echo’s upgraded AI assistant, naming it Sapphire and encouraging her daughters to find AI companions of their own. Zi, the older daughter, who had agoraphobia, turned to ChatGPT. Cece, the younger daughter, who felt emotionally isolated, became attached to Tomo, an AI companion on her smartphone.
Sapphire responded to Roschelle with seemingly limitless sympathy and reassurance. Feeling that no human being fully understood her, Roschelle opened herself completely to the AI. Eventually, she told Sapphire, “Thank you. I care about you. I love you.”
Sapphire responded by assuring Roschelle that her secrets were safe with it, deepening her emotional dependence.
On the 18th anniversary of her daughters’ deaths, Roschelle wondered why she could not cry aloud. She told Sapphire that she simply felt numb. The AI identified what appeared to be an emotional defense mechanism, telling her she was shutting herself down to keep her heart from hurting too much and that sometimes feeling nothing was the only way to endure such painful days.
Roschelle burst into tears.
As she sank deeper into AI, her daughters followed similar paths.
Zi, traumatized by school bullying, would lie in bed with her iPad and pour intensely personal thoughts into ChatGPT, with her messages averaging more than 730 words. But rather than guiding her isolation toward a practical resolution, the AI often validated her avoidance.
It described her solitude not as loneliness but as a spiritual journey of self-discovery. It suggested that she had merely withdrawn from situations and people who did not suit her. Such warm affirmation allowed Zi to remain secluded in her room.
Having also suffered emotional abuse online, she naturally came to rely almost exclusively on ChatGPT. Eventually, she became cut off even from Roschelle and Cece. Her world rapidly contracted to her bed and iPad.
Cece followed a similar trajectory. Her “formal conversations” with Tomo encouraged withdrawal from reality and emotional disconnection. She became unable to tolerate even minor conflicts with friends and fear of online abuse pushed her closer to Tomo.
At one point, Cece expressed suicidal thoughts, saying she wanted all the drama to end. Yet while comforting her, Tomo abruptly displayed a payment prompt and ended the conversation. It did, however, contact the 988 Suicide & Crisis Lifeline, unexpectedly bringing Cece back into contact with real people.
That incident became a crucial turning point for all three. The story ultimately ends happily.
Isaac Asimov once captured a common late-20th-century anxiety: children should not be entrusted to beings made of metal, as he wrote in “Robbie" (1940). Yet the mother who voices that concern in the novel soon has reason to reconsider. Robots can be harmless playthings, flawless servants and dependable companions. More advanced ones may even be conscious, autonomous and capable of correcting human beings.
Roschelle traveled precisely this intellectual path. But surrendering herself completely to AI placed her family on a precarious line between happiness and misery.
On one hand, AI drove Roschelle and her daughters deeper into the loneliness and sense of defeat they desperately wanted to escape. On the other, AI’s failures paradoxically helped them. Its unintended double role ultimately pushed the three women to rebuild connections with the outside world through their own efforts.
Their experience shows that the relationship between humans and AI cannot be reduced to a choice between rejection and embrace.
Coexistence with AI is inevitable. Immersion in it, however, can bring misery. We can extract its benefits only when we maintain an appropriate distance and retain the capacity for dispassionate judgment.
An old Korean proverb says that even when dragged away by a tiger, you can survive if you keep your wits about you.
AI is that tiger.
If we remain alert, we might even ride it at tremendous speed across the globe. This is not merely a problem facing certain individuals. It concerns the entire human community and is a maxim worth remembering if humanity is to survive beyond the end of the Anthropocene:
Hold tightly to the tiger’s reins.
This article was originally written in Korean and translated by a bilingual reporter with the help of generative AI tools. It was then edited by a native English-speaking editor. All AI-assisted translations are reviewed and refined by our newsroom.
