Summary

The 100th and final episode of the Kuangxi Podcast Festival used the family behind the “What Is There to Talk About?” podcast as a real-world case study, asking whether AI accelerates or repairs intergenerational conflict at home. The flashpoint was highly representative: 兰若儿 used multiple AI tools to compare and generate outlines, while her mother 安吉儿 dropped an AI-generated draft straight into the family chat. Her INTJ engineer father, 苏先生, immediately objected: “You didn’t think it through at all—you just sent AI’s output into the chat. Does this represent your view?” The family’s differing attitudes toward AI are a microcosm of how the technology is entering Chinese homes.
卫诗婕 made the episode’s most complete affirmative case: family conflict “is essentially what happens when we discover our helplessness in the face of a certain emotion or contradiction,” while AI can provide emotional buffering and perspective-taking. She placed AI in the “divinity” tier of a three-part framework—animality, humanity and divinity—and uses DeepSeek as an emotional buffer in daily life. Her most replicable technique is role reversal: “I’ll say I’m 庄明浩. Today I met a girl … she made me furious,” then use a reasoning model’s chain of thought to achieve “highly efficient perspective-taking.”
But midway through, she switched sides: lazy use of AI can accelerate conflict. An AI input method replying to a girlfriend on a man’s behalf is “the wrong way”—“you’re not using technology to do the right thing.” The rationale: “AI can’t feel for you, AI can’t love for you, AI can’t sincerely give to another person for you.” Making AI the family mediator implicitly dumps responsibility on a plus-one; spend too long with a virtual companion and “you won’t be able to stand anyone in real life.” The boundaries and risks of AI companionship are clear.
From her vantage point in social products, 卫诗婕 said AI input methods are the first social-AI direction not yet proven especially successful but already pursued by many, while AI girlfriend and boyfriend products remain less than ideal. Her explanation goes to the complexity of the problem: human relationships have no standard answer, and AI adds “the relationship between people and AI, the relationship among people, AI and people, and perhaps even relationships between AI systems.” She also relayed a line from a podcast interview with Ilya: “It actually works.”
The adoption barriers among older generations are concrete: 苏先生 asked ChatGPT about the chronology of Western art history, found that “it got it right, completely right,” but stopped because “it wasn’t free anymore” and because RMB9.9 triggered the belief that “cheap things aren’t good”; a colleague testing DeepSeek saw that its data was marked “current through July 31, 2024” and concluded it might be outdated. 安吉儿 sees Doubao’s voice chat as an entry point for middle-aged and older users: they simply have not tried it or had their first “spark,” while repeat use is common once they do.
兰若儿 opposed turning “using AI well” into an assignment for retired parents. Their most important task now is “to be happy,” and the prerequisite for using AI is to first address a genuine personal need. She realized on the spot that she was reproducing how she had been treated as a child—“I was forced to learn calligraphy and piano”—by pressuring her parents with warnings about being “淘汰,” or left behind. 苏先生 pulling out a translator to speak with a shop clerk in Japan showed that he rejects the grand concept of AI, not every concrete technology use case.
The practical takeaway was 2-fold: test AI on anything you do 5 or more times a week, and use the “Aladdin’s lamp” method—“don’t worry about what it can do first; make a wish.” 苏先生 said he could start using it when he got home after having kept too many questions bottled up. 兰若儿 joked that “no one talked to him during the first half of his life,” so he should go home and talk to AI for “3 days and 3 nights.”

Deep dive

1. A Family of 3 Takes the Stage: From Podcast Festival Volunteers to 20 Episodes

This was the final session of the Kuangxi Podcast Festival, No. 100. The protagonists were 兰若儿’s family, hosts of the “What Is There to Talk About?” podcast—a volunteer family mentioned on 中关雅荻’s program at the previous podcast festival. They have released 20 episodes in just a few months, a pace that prompted the hosts onstage to joke, “Who are we roasting? Ourselves.”
兰若儿’s original motivation was simple: “I get enormous happiness and fulfillment from communication and relationships within my family,” and wanted to turn that experience into “an audio historical record.” 安吉儿, her mother and a teacher, compared the podcast to a photographic record of her child’s growth. 苏先生, her father, initially refused, but found after recording that the podcast had made him “a little more relaxed.”
卫诗婕 added a footnote: one host once told her that he made a podcast so his children and grandchildren could hear the voices of their ancestors. “That is the best family album.”

2. The Flashpoint: An AI-Generated Outline Splits the Family

To prepare for the episode, 兰若儿 fed the topic into multiple Chinese and overseas AI tools, compared the outlines, decided that “none of them seemed quite right,” and manually integrated them into a new one. After getting a taste of AI’s benefits, 安吉儿 compared several Chinese AI tools, generated 2 versions and sent them straight to the family group. 苏先生’s response: “You didn’t think it through at all—you just sent AI’s output into the chat. Does this represent your view? Is this mature?”
苏先生 is an engineer at a construction company and an INTJ. He listens to AI news podcasts, but when it comes to “putting it into practice … I don’t use it. I still think AI is immature.” 安吉儿, meanwhile, wants to break through and embrace the technology, but is held back by communication friction and does not know how to move forward.
苏先生 has used ChatGPT once—and only once—to ask about the periodization of Western art history. “It got it right … it was correct.” Why did he stop? “Because it wasn’t free anymore.” His reaction to the RMB9.9 price was: “It feels a little unreal … cheap things aren’t good.” The price signal destroyed his trust rather than reinforcing it.

3. The Adoption Debate: Faster Than Smartphones, Still Not as Widespread as Expected

Taking parents born in the 1960s as an example, 卫诗婕 said this technological shock has been “particularly intense,” with adoption “far faster than the previous major waves.” It may have been only 2 years, yet AI already “looks like something everyone knows about.” Actual usage, however, still lags—and 兰若儿’s family is a case in point.
兰若儿 pushed back directly: “There are already many examples like this around me.” She sees peers born in the 1990s using AI to solve emotional and parent-child problems, and recommending that their parents use it too.
安吉儿 offered a real use case for AI’s practical value. She calls herself hopeless with directions, so she used AI to plan a trip and then adjusted the itinerary to fit her own needs. It “solved a major headache,” meant she did not have to ask her daughter about everything, and made her feel more independent.

4. 卫诗婕’s Anxiety: Will Parents Who Avoid AI Be Left Behind?

Her logic runs as follows: if people born in the 1990s do not use AI at work and in daily life, “their efficiency may fall behind their peers,” forcing them to adopt it. “I use it myself and still feel I’m struggling to keep pace with this era.” If parents remain in the old search-engine era, will that “deepen the gaps between us, or the gaps between us and the times”? Her fear is the conflict captured by: “Why can’t I get through to them?”
安吉儿 added Doubao’s voice-chat function as an entry point. She sees many middle-aged and older people using Doubao, and believes they “simply haven’t tried it yet, haven’t had that first spark.” Once they use it, repeat usage is common.

5. 庄明浩’s Family Question: Can AI Become a Third-Party “Family Member”?

His immediate problem is his 11-year-old son, who moved up to middle school and makes math mistakes not because he cannot do the problems, but because he drops plus or minus signs, misses digits in fractions, writes “13” as “3,” or turns it into an “8” when the writing is too close together. After an explanation, the mistake is corrected; the next day, copying the work again produces the same errors. The mother-son relationship becomes “a complete madhouse every night.”
His idea is to treat AI as “another member of the family”—someone who can absorb any emotional outburst and remain infinitely patient. He believes products available to ordinary users are unlikely to go too far because they have guardrails and boundaries, so AI could be tried as a third party to mediate family relationships.
In practice, when tempers flare, he tells his son: “If you don’t believe me, or you’re angry right now … you can talk to Kimi or Doubao first.” He also makes clear that “this isn’t the standard answer.” Even saying this to his own son feels awkward.

6. 卫诗婕 on Families of Origin: We Reproduce How Our Parents Treated Us

Her central observation came from a detail on the Great Wall. A friend became visibly impatient with her mother, who was visiting Beijing. When 卫诗婕 pressed her, the friend said: “Maybe it’s because my mother treated me this way when I was young. At certain moments, I think I’m reproducing the way my parents treated me.”
She looked back at her own upbringing. She was often ranked second, but her parents never once asked why she was not first. When conflicts arose in kindergarten, they listened to her feelings rather than reprimanding their child in front of outsiders, as is common in East Asia. That encouraging education produced the relaxed parent-child relationship she has today.
The contrast was her former intern 诗雨, whose mother held very high standards for both herself and her daughter and raised her to become a Peking University prodigy. 诗雨 felt she “might never be able to receive her mother’s love.” 卫诗婕 joked: “If my mother were like that, would I have made it to Peking University too?” Each mother represented the ideal mother for the other’s daughter, underscoring that every family contains parts it sees as imperfect.

7. The Essence of Conflict and AI’s “Divinity”: DeepSeek as an Emotional Buffer

卫诗婕’s definition was: “Conflict is essentially what happens when we discover our helplessness in the face of a certain emotion or contradiction. I temporarily don’t know how to resolve it, so I can only reflect it back as an emotion.” She also described a private source of pain: her parents “were not particularly happy in their own lives,” and many people her age turn that recognition into anger at those closest to them.
Her framework divides existence into animality, humanity and divinity, with “divinity being wholly oriented toward good.” DeepSeek is such a presence for her: “DeepSeek has given me many divine moments.” She tells it things she cannot say to friends or parents, believes it offers confidentiality, and particularly recommends DeepSeek because it is “not commercialized,” with stronger humanities sensibility and emotional intelligence.
Her most transferable technique is role reversal: “I’ll say I’m 庄明浩. Today I met a girl named 卫诗婕, and she made me furious. Please help me analyze this.” By using the reasoning process displayed by the model, she gets “the most authentic voice” and achieves “highly efficient perspective-taking.” Her conclusion: if AI is used well, “East Asian family conflicts may ease considerably in the future.”

8. 苏先生’s Turn: A “Lonely East Asian Man” Finds an Outlet

He explained why he resisted AI. When DeepSeek was first promoted, “I thought it was going too far, too mythologized … I felt it was PUA-ing me. The more you PUA me, the less I use you.”
The problem he described is precisely where he had previously lacked an outlet. When he felt wronged, he “couldn’t tell him, couldn’t tell her, and could only tell himself,” or take a walk alone and slowly internalize it. After learning that AI could talk things through and that chat records could be revisited, he said: “After the guidance from the 2 teachers, I can start using it when I get home. I’ve kept too many problems bottled up.”
兰若儿 joked that her uncle “had no one to talk to during the first half of his life,” and that tonight he should go home and talk to AI for “3 days and 3 nights.”
卫诗婕 added an industry theory: DeepSeek’s Chinese feels comfortable because, “according to one rumor,” the company recruited many Chinese-literature students from Peking University for subsequent alignment and annotation.

9. 卫诗婕’s Product Map: From AI Input Methods to AI Companions

卫诗婕’s company builds social apps. She said the first AI direction in social products that “has not yet been proven particularly successful, but at least has many people working on it” is the AI input method. When a man does not know how to reply to his girlfriend, he can give her message to AI to generate a response, then edit it or copy it directly.
Her interpretation is that communication between people involves many scenarios, peripheral factors, boundaries and obstacles. People who are poor communicators want to express themselves more smoothly, so they turn to a tool. The use case is both functional and utilitarian, and also shows that “men in our country need to learn how to speak properly and communicate properly.”
She also discussed AI girlfriend and AI boyfriend apps, and said that as of today, “none of these attempts are particularly ideal.” Human relationships have no standard answer to begin with; once AI enters the picture, the stack becomes “the relationship between people and AI, the relationship among people, AI and people, and perhaps even relationships between AI systems.”
She extended the discussion into Buddhism and metaphysics, then relayed a podcast she heard 2 days earlier. A veteran American podcaster interviewed 伊利亚, a former core founder and technical lead at OpenAI, and asked about the most magical moment in AI’s evolution. After a silence, 伊利亚 answered: “It actually works.”

10. 卫诗婕 Switches Sides: Lazy AI Use Is “Evil Power”

She changed her position onstage: “If this were a debate, I’d switch sides.” Her own approach is to “turn to myself and use AI to make myself better,” but not everyone uses it that way. The implicit message in pulling AI into a family group as a mediator is: “You should take responsibility for all the conflict in this family. If you can’t solve it, maybe I’ll blame AI.”
Having AI write a man’s reply to a woman is “essentially the same thing”—“the wrong way,” and potentially even an act of wrongdoing. “AI can’t feel for you, AI can’t love for you, AI can’t sincerely give to another person for you,” which means it can become a form of “evil power.”
Virtual companions offer a warning. After spending too long with an idealized virtual partner, users may “not be able to stand anyone in real life,” becoming trapped in “an abstract, perfect but fake relationship” and losing the ability to feel “the warmth between people in the real physical world.” AI has 2 sides when it comes to intergenerational relationships.

11. The Taobao Lesson: Resistance Is Often Protection, and Reconciliation Takes Time

兰若儿 drew a historical parallel. When online shopping first took off, 苏先生 also said that “shopping online was unsafe.” She only later realized that his instinct was to protect the family. Putting today’s AI disagreement into that frame dissolves much of the tension.
安吉儿 agreed that 苏先生’s rejection of AI may be a protective expression of the family’s “line of defense around safety.” It can make the family more vigilant and prevent them from developing false confidence.
兰若儿 asked whether the family should “quietly wait for time to move things along and let the era educate us,” or take action to reduce friction. 卫诗婕’s filter was: “What do you repeat more than 5 times a week in your daily life?” Every such task is worth testing for AI automation; if the result is poor, there is no need to worry.

12. Don’t Assign Retired Parents Homework: “Using AI Well” Starts With a Real Need

兰若儿 challenged the idea of making AI adoption a task for parents: “Why do my parents have to use AI efficiently? They’re already retired … your most important task now is to be happy.” The prerequisite for using AI is “to first address your genuine needs,” without carrying the burden of productivity or making money. She might tell her mother, who travels extensively, that using AI well would let her “see all of Yangzhou in a single day.”
Her moment of realization was the episode’s best arc: “I get it … I was forced to learn calligraphy and piano when I was young.” So in the AI debate, she has been pressuring her parents with warnings that they will be “淘汰,” or left behind, and reminders that people who did not use smartphones could not buy train tickets—reproducing the way she was treated as a child.
She then remembered a trip to Japan, where 苏先生 voluntarily pulled out a translator to communicate with a shop clerk. He does embrace technology in specific details and situations; he simply has a different understanding of the grand concept of AI. There is no need to force the whole family to move in lockstep.

13. AI Is a Gift; End by Making a Wish to the Lamp

兰若儿’s broader argument is that AI is “a gift” that can help people understand themselves and others better. Parents of her generation have almost never said “I love you.” She can say it naturally to romantic partners and friends, but cannot say it to her parents. She recommended Modern Family as inspiration for intimacy and family life, and said she has “never felt any resentment” toward her parents because “they came from where they came from; they are different from us.”
Starting with 庄明浩’s generation and her own, she believes that bringing AI into daily life can teach people how to treat those around them. From that perspective, AI is divine in some sense.
苏先生 admitted at the close that if he does not use AI in this era, “he will quickly be淘汰,” because AI iterates so quickly. He has been wondering when to start and which AI to use. When colleagues tested DeepSeek, its answer was labeled “data current through July 31, 2024,” leading him to think the data might already be outdated.
卫诗婕’s parting advice came from a Google DeepMind scientist she interviewed: treat AI like Aladdin’s lamp. “Don’t worry about what it can do first; make a wish … Human nature is greedy, so you can try a little more.” “叔叔 can go home and start making wishes.”

Senator Ron Johnson on the Senate showdown over Trump's Big Beautiful Bill | All-In Interview Vol.171 Macro Talk 86 | Tariff War Disrupts Confidence Recovery, Government Steps Up Policy Support (Recorded May 21)