Summary

Workplace AI is becoming an assumed input, shifting the competitive question from whether employees use it to whether their output is good. Kevin says managers should presume adoption “unless specified otherwise,” while Casey argues bad AI-assisted work should be challenged as bad work—not as illicit tool use. Ed’s caveat matters: output-only evaluation may suit production code, but education “is not a widget-making” exercise.
Generative AI’s deepest creative risk may be eroding the incentive to learn and make things, not merely eliminating jobs. Ed is “stunned” by ChatGPT’s creative facility and fears a world where “the need for human creativity disappears.” As the technology becomes more powerful, intoxicating and affirming, people may question both the point of making art and the value of studying an art form.
Generative AI is already redirecting consumer behavior and weakening businesses built around older discovery channels. Ed now uses ChatGPT as his “new Google” for research and general questions, while noting that Google’s own AI answers mean fewer people visit links, unsettling industries dependent on search traffic. Kevin separately says he uses AI to help think through difficult, emotionally loaded emails.
The broader consumer-tech problem is compulsory dependence on products users increasingly suspect are harmful. Ed compares device usage to cigarettes in the 1990s, except phones mediate banking and shopping: “the only way to do your banking…is to smoke your cigarettes.” That integration makes withdrawal unrealistic and leaves him feeling he uses his phone too much.
Digital boundaries work best when refusal is direct and, where possible, paired with a usable alternative. For grandparents posting children publicly, the group recommends a direct conversation plus a shared album, WhatsApp thread or family text. Calendar advice similarly emphasizes understanding the other person’s norms and technical constraints before asserting a preferred workflow; public audio calls for a plain request to use headphones or lower the volume.
Engaging scammers offers little upside and can increase both human and technical risk. Many operators may be trafficked or coerced, while any reply can mark the recipient as a “warm phone number” for more targeting. For a listener who found cryptocurrency by guessing private keys, the answer is equally direct—“stay out of the mess”—and route any vulnerability through a bug-bounty program, if the project has one, rather than taking funds.
AI anxiety becomes constructive only when uncertainty and human agency remain in the frame. Ed recommends therapy when fear becomes isolating; Casey says citizens should still discuss concrete harms; and Kevin warns that high-P(doom) conversations can leave people feeling “we’re all screwed.” Conditional paths—good decisions can improve outcomes, bad ones can worsen them—prepare people without presenting catastrophe as destiny.

Deep dive

1. Historical tech failures begin with human incentives, not broken machinery

Helms’s favorite technology snafu came from the 1950s Cold War: planners considered firing a nuclear warhead at the moon to terrify the Soviets. Research suggested it could miss, slingshot around the moon’s gravitational field and return to Earth—which, he joked, “frankly would have served us right.” Time, money and resources were spent before the plan was abandoned.
His lighter specimen was Elk Cloner, a computer virus that popped a programmer’s “dumb poem” onto computers. Compared with modern attacks focused on stealing millions in Bitcoin, the group’s nostalgic verdict was: “Bring back fun hacks”—the cheeky kind that makes victims chuckle rather than drains their accounts.
Helms once felt “ahead of the curve,” buying new laptops and adopting the iPhone early. Over the past five or ten years, technology has “blown past” him: upgraded apps assume fluency in an unfamiliar visual language, like entering a grocery store where not only the shelves moved but “the broccoli’s blue.”
His macro snafu is device dependence itself. Phones feel like cigarettes in the 1990s—widely understood as harmful while companies keep improving and marketing them—but with a crucial difference: banking and shopping now require the device. The joking policy prescription captured the dilemma: permit Instagram only outdoors, 15 feet from the door.

2. Generative AI threatens the motive to create, not only creative jobs

Helms uses ChatGPT for research and general questions; it has become his “new Google.” He still searches conventionally, but says Google’s AI means people no longer visit links, threatening industries built around search referrals. Kevin separately says he uses AI to help think through emotionally loaded emails.
Technology has always been integral to filmmaking, from photography and camera movement to digital image manipulation and animation. ChatGPT nevertheless introduces “a completely different paradigm”: Helms suspects some executive project notes are thinly veiled AI output and is simultaneously “stunned” by the system’s facility with creative assignments.
The creative community is “terrified and for good reason,” but Helms’s central fear is human rather than commercial: “When the need for human creativity disappears, like, will we still be creative?” If machines consistently produce better work, people may question both the point of making art and the value of studying an art form.
His current position remains unsettled. AI can be a useful tool or creative partner while some creative work stays reserved for the person—but the products will become “more intoxicating to use,” more exciting and more affirming. That escalating appeal makes today’s personal boundary inherently unstable.

3. Digital boundaries work best when refusal comes with an alternative

Asked how to stop a future grandparent from posting children on Facebook, Helms favored a direct conversation; he posts no pictures of his own children publicly. Kevin added the practical complement: create a shared photo album, WhatsApp thread or group text so relatives receive an abundant private stream rather than feeling deprived.
Calendar etiquette exposed a genuine disagreement. Casey’s “old school” norm is that people agree on a time and manage their own calendars; Kevin would simply send every invitation. Their advice was to confirm that the friend understands a shared event does not reveal the full calendar, then account for organizational rules and “calendar culture” before insisting on one workflow.
Public phone audio calls for a polite request, though the speakers admitted enforcement is awkward. Casey’s fallback was social friction: ask detailed questions about the game, score or basketball draft until the person gets annoyed. The absurdity is sharper when earbuds may cost $15, yet strangers increasingly treat the whole subway or airplane as their speaker system.

4. The safest response to online bad actors is disengagement

A listener enjoyed taunting scam callers but worried that some worked under trafficking and extortion. Kevin’s answer was “cold-ignore”: engagement creates unknown risk because the caller may possess personal information. Casey added that replying—even abusively—can identify a live, “warm” number and generate still more scam traffic.
The moral argument went beyond self-protection. Mocking somebody trapped in a scam compound makes an already difficult life harder and reinforces “not your best self.” Hyperconvincing AI scammers might worsen fraud, but Casey floated one qualified upside: automating that labor could become a human-rights victory if coerced workers were released.
A second listener had guessed weak blockchain private keys around 2016 and found an address holding several thousand dollars in cryptocurrency. Casey’s verdict—“stay out of the mess”—compared taking it to joining a mob meeting because it presents a moral quandary. Kevin’s lawful alternative was reporting the weakness through the project’s bug-bounty program, potentially earning a reward without stealing funds.

5. Workplace AI normalizes faster than institutions can define competence

A team lead secretly used AI for brainstorming, then recognized the same suggested solutions in a junior developer’s work. Helms called the manager’s impulse to confront the employee hypocritical and proposed greater transparency. Casey reframed the test: if the suggestion is bad, explain why it fails; if it works, AI use itself may not be the problem.
Kevin’s operating assumption is that people will use AI in almost every job. Citing economist Tyler Cowen’s classroom policy at George Mason University, he described requiring students to use a chatbot and grading the finished result. For a software team, the analogous standard is whether the code works and compiles—not which permitted tool produced it.
Helms preserved the critical distinction: education “is not a widget-making” exercise. A polished submission may not show whether a student understands the subject, and outsourcing thought can weaken the critical and creative capacities school exists to develop. His darker long-term question was, “What’s the point of learning to do it?” if AI will always do it better.
A NASA scientist’s colleagues framed AI abstinence as superior intelligence and called unaided problem-solving “the whole fun of being alive.” Helms suspected some critics quietly use AI; Kevin saw fear that machines might outperform supposedly irreplaceable expertise. Kevin recommended kindness toward people who use AI well, while Casey offered an abacus to anyone rejecting calculators for “cognitive stimulus.”

6. AI disagreement is a relationship problem before it is a persuasion problem

A listener’s girlfriend had a visceral aversion to AI despite his personal and professional enthusiasm. Helms treated that mismatch as surmountable: every couple contains interests one partner does not share. Casey’s broader rule was to “distribute the weight” of obsessions across friends, clubs and communities instead of making one partner absorb everything.
Kevin’s concrete route to organic interest was usefulness. Identify something the skeptical partner values or struggles with, test whether an AI model can help, and show the result without forcing it on them. People rarely believe a technology matters personally until it solves a problem they actually have.
The limit is consent. Ed advised taking no for an answer and cooling the AI discussion; Kevin also suggested finding an AI club or another outlet for these conversations. AI may eventually enter the skeptical partner’s life in ways that make her more open to it, but the relationship need not become a referendum on the technology.

7. Doom talk needs uncertainty, agency and somewhere safe to land

A tech-avoidant listener felt existentially isolated by grim AI forecasts and feared frightening loved ones. Helms recommended therapy as a place to process the burden without transferring all of it to friends and family. “None of us really know what’s coming,” he stressed, and AI optimists “maybe…are right,” making categorical Chicken Little predictions premature.
Casey resisted silence, arguing that democratic life requires discussing concerning technologies. His prescription was specificity in small doses: talk about sycophantic chatbots or users developing messianic beliefs rather than announcing inevitable doom. People are “allowed to be a bummer at least 10 or 20% of the time.”
Kevin has found that his gloomiest, highest-P(doom) conversations go badly because listeners leave believing “we’re all screwed” and powerless. More optimistic discussions preserve agency: if people make good decisions, outcomes might improve; if they do not, outcomes might deteriorate. The essential move is replacing inevitability with choices and branching paths.
Helms’s final dilemma concerned tech support for his 85-year-old mother. Kevin now favors simply fixing parents’ devices rather than making every repair a lesson. Casey would still nurture genuine curiosity: his mother used Claude to build a playlist for her 50th wedding anniversary party, then noticed it was “way too nice.” Fixing the problem remains “an act of love.”

AI Investor Panel: Where Smart Money Is Actually Going in AI | EP 219 Costco (Audio)