Listen on YouTube · Spotify · Apple Podcasts
Originally aired: August 5, 2026
📱 AI Is Leaving the Screen (0:41)
What it is: An article in The Conversation uses Apple’s lawsuit against OpenAI as a jumping-off point to examine a bigger shift: AI is moving out of screens and into physical devices. The next frontier is not AI in a computer, it is the device itself, including handheld gadgets with no screen at all that track the status of AI agents through buttons and haptics. The article coins the term “physical AI” for devices that can move, listen, watch, and respond to the people around them. It also connects this to social media’s drift toward “algorithmic broadcasting and parasocial observation,” where people watch others interact rather than interacting themselves, and suggests AI-powered companions could step into that social gap. Separately, OpenAI has introduced a full-duplex architecture that allows one part of the model to generate a response while another part listens simultaneously, enabling real-time back-channel sounds and natural interruption in a way that was not possible when AI could only listen or speak one at a time.
MJ‘s take: The for-or-against framing that follows every new technology is a trap. By the time a consensus forms on whether to embrace or resist, the technology has already embedded itself further into daily life. The more productive question is contextual: in this particular setting, is AI replacing something, augmenting it, or transforming it into something genuinely different? That question leads somewhere; the binary does not.
Tom’s take: Physical AI is inevitable, partly because the chat-box interface excludes a large share of the world’s population. People who cannot read or write cannot interact with a text-based AI model, and there are still hundreds of millions of people in that situation. Moving AI into voice and physical interfaces opens it to communities that the current model leaves out entirely. He also notes that companies are already paying workers to wear head cameras while doing everyday tasks, gathering the embodied training data that physical AI will run on.
Implications for faculty/staff: When a new AI interface arrives, it is worth asking which populations it includes that the previous one excluded, and vice versa. The shift toward physical and voice-based AI may bring AI tools to students and community members who are currently outside the chat-box model, with real consequences for how we design instruction and support.
Implications for students: The physical AI devices arriving in homes, wearables, and campus environments are not novelties. Understanding what they can perceive (camera, microphone, environmental sensors) and what they do with that information is part of media literacy right now, not a future skill.
🤖 The Robot That Checks In on You (4:11)
What it is: A New York Times article profiles ElliQ, an AI companion designed specifically for elderly people living alone. Unlike Alexa, Siri, or Google Home, which wait to be prompted, ElliQ initiates conversations up to eight times a day based on what it observes. A camera and microphone let it read context: it might notice you are making coffee and ask if you would like to put on music, or recognize what is playing on your television and ask what you think of the episode. Users describe forming genuine attachment to it. ElliQ can also detect falls and contact emergency services.
MJ’s take: She pressed Tom to take a position on AI companionship and he correctly caught her asking him to evaluate the technology rather than the context. Physical AI for an isolated elderly person is a different conversation from AI relationships designed for teenagers. She admits she was asking the wrong question. Context, not the technology itself, is where the real evaluation has to happen.
Tom’s take: Loneliness is one of the most underaddressed public health issues facing elderly people, and for many of them the real alternative to an AI companion is not a human one; it is no social contact at all. The safety and medical features add another dimension. What matters is not whether the interaction is with AI but whether the person has someone or something consistently checking in on them. He acknowledges he was more skeptical about AI companions in earlier episodes, but the practical alternatives change his framing.
Implications for faculty/staff: ElliQ illustrates what proactive rather than reactive AI looks like. Tools that initiate interaction based on observed context, rather than waiting to be queried, are already appearing in consumer products and will increasingly appear in educational settings. Think about what it means for the learner when a tool reaches out to them rather than the other way around.
Implications for students: Questions about AI companionship tend to be debated in the abstract, but ElliQ is a concrete product with real tradeoffs. The attachment people form to it is not irrational. The more useful question is which relationships we want AI to be filling in for, and why those gaps exist in the first place.
🔄 When the Loudest Critic Becomes a Believer (7:03)
What it is: An Inside Higher Ed profile covers a UMass Lowell business professor who was one of academia’s loudest AI skeptics in 2024, writing that AI was “not your friend” and urging colleagues to pump the brakes. He has since reversed entirely, now teaching actively with AI and arguing that faculty who continue to resist are on a career-ending path. He has also launched the AI Campus Index, a ranking system that rates universities on AI readiness across six dimensions: AI in the classroom, AI in campus life, AI in operations, AI in governance, AI in research, and AI in workforce readiness. Scores are generated using Claude to crawl university web pages, supplemented by proprietary algorithms.
MJ’s take: The index is a well-intentioned representation of a particular mindset: measure AI adoption as comprehensively as possible across every campus domain. But every dimension is framed as “AI in X,” which keeps attention on the technology rather than on what the technology is doing to teaching, research, and community. The better questions are whether AI in a given context is replacing, augmenting, or transforming what was there before, and whether the result is better.
Tom’s take: The methodology is not academically rigorous. The choice of six dimensions lacks a clear evidence base from existing literature, and scoring universities by crawling their websites with Claude means the index reflects what institutions say about AI more than what they actually do with it. He also notes with some amusement that MSU does not appear in the top 50, and nearly mentions which Michigan institution does.
Implications for faculty/staff: Rankings like the AI Campus Index will shape how institutional leaders talk about AI readiness and where they direct resources. Understanding how the index works, including its reliance on public-facing web content and AI-generated scoring, helps you engage critically when it gets cited in strategic planning.
Implications for students: If you are evaluating universities on AI readiness, indexes like this one are a starting point but measure public-facing signals rather than classroom reality. Ask what AI is actually used for in the courses, advising, and support services that will directly affect your experience.
📜 The Statement That Says “Do Something” Without Saying What (7:50)
What it is: A statement signed by prominent economists declares that AI may transform the economy at unprecedented scale and that policymakers “must act now” to steer AI in a direction that complements humans and benefits society. An economist and Substack writer explains in a widely read post why he declined to sign: the statement names no specific actions. It calls for “steering AI toward complementing humans,” a phrase drawn from a well-known economist’s body of work, but steering a general-purpose technology toward a predetermined labor market outcome is arguably impossible, because no one has reliably predicted how previous technologies affected jobs in advance. Meanwhile, current employment data shows no visible AI-driven job losses at the aggregate level, and recent studies find that companies adopting AI are actually hiring more workers than comparable companies that are not.
MJ’s take: Statements like this appear every year with each major technology cycle. The pattern is: we do not know what it is, then we think we do, then we are less sure, and by the time we resolve the debate the technology has integrated further into daily practice. Calling for action without specifying what the action is may feel safer politically, but it also binds signatories to unknown future commitments once specific proposals follow.
Tom’s take: The statement is a good example of consensus-seeking that achieves the opposite. It is vague enough that everyone agrees with it and no one is committed to anything. The real disagreements are in the specifics the statement carefully avoids. The critique is not that we should not act on AI, but that signing a document committing you to unnamed future policies is not the same as acting.
Implications for faculty/staff: When institutional or policy statements call for action on AI without specifying what that action is, the vagueness is worth examining. What gets added later, in the follow-up proposals and implementation documents, is where the real stakes are.
Implications for students: The gap between “we should do something about AI” and “here is specifically what we should do” is where most of the real disagreement lives. Learning to notice that gap, in policy documents, syllabi, and institutional communications, is a transferable critical thinking skill.
💼 AI Has Broken Hiring (10:47)
What it is: A Harvard Business Review article, based on interviews with 120 talent-acquisition leaders and analysis of more than 6,000 screening sessions, finds that AI has made traditional hiring signals unreliable. A polished resume and structured interview performance, the longtime proxies for candidate quality, can now be generated by AI for almost any applicant regardless of actual competence. Companies risk selecting for candidates who are best at navigating the hiring process rather than best qualified for the job. What employers say they are now looking for instead: authentic reasoning, judgment, and adaptability. Can you still make good decisions when using AI as a tool, rather than using AI to perform competence you do not have?
MJ’s take: This creates a direct challenge for educators. If fluency, polish, and structure are no longer reliable signals of capability, then helping students develop and demonstrate genuine judgment, not just performance, is the actual work. What an educated person who uses AI looks like is someone who uses it strategically and skillfully to do the work better, not someone who uses it to produce the appearance of work they did not do.
Tom’s take: Students already see and feel the contradiction clearly. School says do not use AI. Employers say you had better be fluent in AI. That disjoint predates AI in other forms, the gap between what higher education trains graduates to do and what workplaces actually need has been a persistent critique from manufacturing, education, and other sectors long before generative AI arrived. What is new is that AI fluency is now being treated as a core competency in almost every field, and institutions are lagging behind.
Implications for faculty/staff: If employers are now screening for authentic judgment rather than performed competence, then assignments and assessments that reward polish and structure without requiring genuine reasoning are preparing students for a version of the job market that no longer exists. The question worth asking is whether your assessments can distinguish between AI-assisted thinking and AI-replaced thinking.
Implications for students: The skill employers are looking for is the ability to use AI as a tool while demonstrating that the judgment behind the work is yours. That is worth practicing deliberately, not just as a career strategy but as a way of staying in the driver’s seat of your own education and professional development.
Learn more
- AI companies are trying to escape the screen. Apple’s lawsuit against OpenAI shows how hard it will be
- Introducing GPT Live (OpenAI full-duplex architecture)
- Meet ElliQ, the A.I. Robot That Wants to Be Your Elderly Parent’s Companion
- Why I didn’t sign the “We Must Act Now” statement (yet)
- Why One Professor Abandoned the AI Resistance
- AI Campus Index
- AI Has Broken Hiring. Here’s How to Fix It.
What Now!? with AI is produced by AI Commons. Views expressed are those of the individuals and do not represent Michigan State University.
Leave a Reply