@vivekmurthy 9/ Julia's small signal that others are missing: AI companionship is the next social media. The horror stories we're seeing now aren't trickling down because they seem too extreme for regular people to relate to. We're ignoring the problem until it's too late.

@vivekmurthy 8/ The practical guidance for educators is telling: don't just demand better information from AI tools—demand more connections. Bake social connection into your requirements.

@vivekmurthy 7/ Julia's optimistic about AI democratizing social capital—giving everyone access to insider knowledge and networks.

@vivekmurthy 6/ Auditing for isolation: we should audit AI tools for isolation the way we audit for bias. Usage patterns could reveal when tools are perpetuating loneliness rather than alleviating it.

5/ Her solution framework is surprisingly concrete: Massive investment in social infrastructure (we've ignored @VivekMurthy's warnings for too long) AI tools should actively teach help-seeking and encourage offline connection Business model innovation around connection brokering

4/ The research is mixed but concerning. Heavy ChatGPT users are more likely to be lonely—though we don't know causation. One study found AI companion users report social benefits, but that could just be rationalization or short-term thinking.

3/ We have millions of people who "want more connection than what they have access to right now"—literally the definition of loneliness. AI companions are perfectly positioned to exploit this gap in the market.

2/ Quick primer on disruptive innovations: they start at the bottom of the market, compete on access rather than quality, and get ignored until it's too late. AI companions fit this pattern perfectly—they're targeting our loneliness epidemic.

9/ The core tension: AI can democratize access to information and support, but relationships aren't just about information transfer. They're about cultural capital, mentorship, and the messy process of human development that can't be easily replicated by algorithms.

8/ Her solution pathway: experiential learning must become a "market imperative, not a pedagogical choice." But she's skeptical about whether market pressure will actually force educational institutions to change fast enough.

7/ The early career ladder problem is severe. AI is displacing entry-level jobs disproportionately, which means fewer opportunities for mentorship and "learning how to work." Julia's assessment: "We're really screwed on this front."

6/ For parents wondering what kids should learn: Julia chose her daughter's school for critical thinking, public speaking (starting in kindergarten!), and opportunities to practice "the inherent friction and messiness of human relationship."

5/ Julia's contrarian take: she's "not super interested in skill building without relationship building." This isn't just pedagogical preference—it's about economic reality. Skills without networks are an incomplete formula for career success.

4/ The AI angle is counterintuitive. Yes, AI can provide on-demand, personalized support. But if we're not careful, it might atrophy our ability to seek help from actual humans. We risk optimizing for convenience while weakening essential social muscles.

3/ Low-income and first-gen students use writing centers and tutoring the least, despite needing them most. They experience help-seeking as evidence of failure, not as a strategic skill. Social class shapes how we approach seeking help.

2/ Julia's POV on help-seeking behaviors: creating support systems for students is important, but we need a stronger focus on teaching students the confidence and skills to actually seek help in the first place.

@davidblake @degreed @Maestro His prediction he'll be very right or very wrong about: "I'm going to send my children to a university of my making." It's either Dad's university or no university at all. 7/8

@davidblake @degreed His company built @Maestro - an AI tutor that runs on your desktop, not in a browser. Key insight: Most AI tools help you do tasks better, but don't make you smarter. Maestro is designed with "friction" - you need to struggle, get feedback, and practice to actually learn. 6/8

@davidblake @degreed The early career ladder problem: Entry-level roles are simultaneously the easiest for AI to displace AND the most helped by AI. But employers are too focused right now on "How do I reduce my current workforce?" and not "How do I rebuild talent pipelines?" 5/8

@davidblake @degreed Blake's prediction: AI accelerates the labor vs capital divide that's been happening for 30 years. His advice to his own kids? "You won't want to pay your rent with your labor in the future. Whatever your dream - chef, pilot - you need to own the restaurant, own the plane." 4/8