Model Behavior
Is AI creating the faces, voices, and celebrities of the future?
AI-generated personas are landing record deals, brand sponsorships, and modeling gigs. AI podcast startups now produce thousands of episodes per week at near-zero cost. Agencies like Xicoia are launching rosters of “hyperreal” digital stars. Artificial “talent” isn’t sophisticated enough yet to fool discerning audiences, but that doesn’t make them harmless. What they could do is shift cultural expectations more toward a slop of sameness: scalable, algorithm-friendly output increasingly divorced from craft, experimentation, and soul.
Even worse, could a critical mass of slop replace human artists? Many are already underpaid and underrepresented, and the viability of these artificial personas may even fall along economic lines, further damaging the worlds of art and craftsmanship.
We’ve seen labeling AI-generated art meaningfully influence how it’s perceived: studies show audiences consistently rate “human” art as more valuable than identical works labeled as machine-made. The line between helpful and dystopian AI is also felt deeply, if the recent backlash against Friend AI is any indication. People still value human creation and interaction; any new entrants into the AI future would do well to build with that in mind. Labeling AI clearly, treating it as opt-in on creative platforms, and elevating human-driven labor rather than quietly replacing it aren’t propping up dying art forms. They’re helping to preserve the value in our ability to create.
Superintelligence, Sponsored™️
AI was supposed to help people find clearer, smarter answers. Now, it’s helping them shop. Amazon is planning to insert ads into Alexa+ responses. Perplexity is testing affiliate-style link insertions. Mattel is working with OpenAI on AI-enhanced branded storytelling and product development. Will it all pay off?
People treat AI like a helpful agent, not a marketing vehicle. The more the end-to-end advertising experience on AI platforms feels low quality or inhuman, the likelier the backlash. Skechers got heat for some terrible gen AI ads, and the viral Friend AI campaign on the New York City subway led to mockery, vandalism, and accusations of manipulation and creepiness.
The traditional web experience is already degraded. Search is overrun with SEO spam and affiliate-riddled content. Irrelevant ads are still everywhere. AI shouldn’t just replicate those problems in a new format. It should be something better — and it can be.
Winners & Losers
AI could expand educational access in underfunded and rural communities — especially if it helps support overworked teachers and unlock new forms of differentiated learning. Tools like NotebookLM’s updates and Claude in classrooms show what’s possible when AI works with educators, not instead of them. AI as an assistant to better-trained, better-supported human educators could drive a “Teach for America”-style movement with AI fluency at its core.
The need is there, not just because of access gaps, but because of early evidence that over-reliance on AI reduces problem-solving ability. A generation of students trained to always ask chatbots designed to please them may lose valuable critical thinking skills. These effects are already measurable, and uneven access to AI tools could reinforce the divides between public and private education, or urban and rural systems. Layer that with existing divides and rates of higher education, and you have a recipe for some serious problems.
Anthropic’s Economic Index shows large geographic and sectoral divides in AI impact. Lower-income users may be less shielded from misinformation, less likely to have access to training, and more likely to be influenced by AI output. AI literacy must be foundational, not as a privilege for advanced programs, but as a built-in layer of modern education.
Speaking of literacy: corporate AI transformations are failing. It’s not because the tools are bad, but because nobody knows what to do with them. A recent MIT study found 95% of enterprise GenAI pilots falling short. Executives don’t understand what’s possible, employees don’t have the space to experiment, and internal training programs are either outdated or not up to par.
The companies getting it right aren’t chasing the hype. They’re restructuring operations: automating grunt work, rethinking service roles, and giving people tools to build, test, and optimize internal processes, things AI is already quite good at.
This is where the AI “bubble” may burst: not with the models, but with the companies that can’t integrate them and the inevitable impact on bloated company valuations when AI’s flashiest promises give way to something more mundane. Some companies will feel fast, fluid, and future-aligned. The others will feel like relics.
AI continues to evolve faster than institutions, industries, and imaginations can keep up with. If we want it to reflect the best of us, we have to keep it from inheriting the worst of us. Is new technology really going to change the world if we keep most of the world from its high-potential applications? Do we want even the art of selling to be devoid of humanity? The choices we make today will determine our lives, and those of future generations, sooner than we might realize.
Creative ideation, integrated learning, and high-touch service jobs are still irreplaceable. Meanwhile, administrative friction, duplicative documentation, and meaningless reporting could be automated away. I’m reminded of David Graeber’s Bullshit Jobs and how it defines the many useless roles in modern corporations. Can’t AI free us from those and make more room for us to do the fun stuff?
If this was helpful, consider becoming a paid subscriber. Your support helps me keep the work independent and exploratory.
Thanks for reading.
Reading List from Today’s Update
AI artist Xania Monet signs record deal — Billboard
AI podcast startup plans 5,000+ shows a week — The Hollywood Reporter
Xicoia launches “hyperreal” AI talent studio — Deadline
Bias against AI art: Why labeling matters — Columbia Business School
Amazon CEO wants to put ads in Alexa+ conversations — TechCrunch
Perplexity affiliate model rollout — Perplexity / X
Skechers pulls AI influencer ads after backlash — Creative Bloq
Google launches NotebookLM video and studio features — Google Blog
Anthropic education report: How educators use Claude — Anthropic
Anthropic Economic Index (September 2025) — Anthropic
MIT report: 95% of generative AI pilots are failing — Yahoo Finance