AI's Privilege Expansion
Weekly writing about how technology and people intersect. By day, I’m building Daybreak to partner with early-stage founders. By night, I’m writing Digital Native about market trends and startup opportunities.
If you haven’t subscribed, join 60,000+ weekly readers by subscribing here:
AI's Privilege Expansion
Last week, I was on a hike with my brother when we came across two paths on the trail. Naturally, my brother made a joke about Robert Frost’s famous poem, “The Road Not Taken.”
“Did you know,” I said to my brother, “that ‘The Road Not Taken’ is actually widely misunderstood?”
He responded that he didn’t know, asking me to explain. Unfortunately I’d already exhausted the extent of my knowledge—oops. I remembered that “The Road Not Taken” is misinterpreted (most people even get the title wrong, calling it “The Road Less Traveled”), but I couldn’t remember exactly why the poem is misinterpreted.
So, like any good Millennial, I turned to Google.
But after a few minutes of searching, I still couldn’t find a good answer. Most Google results said the same thing—that Frost’s poem is about individualism, about the importance of taking the less-trodden path in life. But I knew that was wrong. (The poem’s narrator even admits that the two paths “equally lay / In leaves” and “the passing there / Had worn them really about the same.” In other words, one path isn’t less traveled.)
Eventually, I gave up on Google. Instead, I turned to ChatGPT. And immediately, I got my answer:
Basically, a careful reading of “The Road Not Taken” reveals that Frost isn’t making a statement about individualism; rather, he’s making a statement about our self-deception when constructing the stories of our lives. We tell ourselves that our decisions make all the difference—but this happens in hindsight, and we don’t truly know. The poem is about choice and fate.
Anyway, enough of the AP Lit analysis.
The point is: Google wasn’t particularly helpful in figuring this out; an AI chatbot, meanwhile, gave me the answer in seconds. This experience made me think of a phrase that I saw on Twitter from my friend Warren Shaeffer: “Privilege Expansion.”
Warren argues that Privilege Expansion is when technology expands access to a good or a service beyond what was previously available.
In the case of Frost’s poem, I essentially got my own personal English tutor—for free. Someone might spend thousands of dollars on a college-level literature course to gain that same deep reading of Frost’s work. An LLM gives me a lit professor in my pocket.
Privilege Expansion: Past Technologies
Of course, the internet already gave us Privilege Expansion to some degree.
Even around information like Frost’s poem, the internet expanded access—enough searching, and I probably could’ve found my answer. (I ultimately found a good breakdown of “The Road Not Taken” from The Paris Review.)
Technology innovations like the internet and mobile improved access to things that were previously (1) expensive, (2) inconvenient, or (3) completely out of reach. The example Warren gives is Uber: with Uber, everyone could have their own black car driver in their pocket. Other examples might be online tutors (e.g., Preply), therapists (e.g., Headway), or medical information (e.g., WebMD). Services that were previously hard to get—or expensive to get—became more available to us.
We’ve also seen Privilege Expansion in hardware: we all have smart TVs mounted on our walls; today, those TVs cost a fraction of what they cost a decade ago.
Or think of the supercomputer in your pocket. We’re not far removed from flip phones and grainy, pixelated phone cameras—but now, we have devices capable of shooting entire feature films. The director of the popular zombie franchise 28 Days Later just revealed that his new installment, 28 Years Later, was shot entirely on an iPhone—by far the biggest movie to date filmed on an iPhone. The crew outfitted the iPhone with an aluminum cage that allowed for different lenses to be attached to the phone (see image below), now uniquely possible because the iPhone 15 is the first iPhone to be able to shoot 4K “log” files, which have a lot of flexibility to be edited in post-production.
Simply put, a $75M movie was shot on the device you have in your pocket. Pretty cool.
Technology’s ability to broaden access is a recurrent theme on Digital Native—we often talk about how Oscar-winning films like Parasite are edited on Apple’s Final Cut Pro software, which I can buy for $299. Over time, technology—both hardware and software—steadily make previously inaccessible goods and services newly accessible. That’s the broad arc of progress.
AI, of course, is the new catalyst here. The internet and mobile brought step-changes in Privilege Expansion; hardware innovations like the iPhone did too. But AI accelerates everything by removing the need for the most expensive and inaccessible remaining variable—the human.
Use Cases for Privilege Expansion
A good formula for AI’s Privilege Expansion might be:
Expensive Human-Centric Service + AI = Better Access & Affordability
Let’s revisit this chart, one of the most-cited in Digital Native. It tracks the price change in consumer goods and services over the 25 years:
As we saw above, products like TVs are at the bottom of the chart, dramatically more accessible to us. The top half of the chart, meanwhile, is filled with human-centric, hard-to-get services. These are the services that are subject to change under AI’s Privilege Expansion.
Tutors & Teachers
Education delivery is a natural fit for AI disruption, because the barrier has always been the human.
In July’s How AI Will Change Education, we looked at how student-to-teacher ratios have trended over time. America’s ratio has been falling—a rare bright-spot in education, as costs balloon and test scores suffer—and we now hover around 15 students for every one teacher.
Of course, this varies by state: a student in Maine gets a lot more attention than a student in California.
And naturally, student-to-teacher ratios vary by income. Lower-income schools can’t afford generous ratios, with fewer (low-paid) teachers spread thinly across large student bodies. This is why socioeconomic status is such a strong predictor of educational attainment.
Of course, a 1:15 student:teacher ratio still isn’t great—and it’s certainly not 1:1, which is where AI comes in. The teacher is the bottleneck, and AI can act as teacher. This effectively gets us to the Holy Grail of education, the 1:1 student:teacher ratio. With AI, every student can get an affordable, personalized learning path. Technology can’t replace the human-level engagement of a good teacher or tutor, of course—but generative AI can get us closer than past technology breakthroughs. The biggest education companies built this decade, in my mind, will be personalized learning and tutoring businesses.
Think of the example earlier with Robert Frost. That kind of analysis might previously be limited to the expensive college literature course; now, everyone can access it through a chat interface. And that was with ChatGPT, not a specialized tutoring product personalized to you.
This is also an exciting opportunity for equality: high-income students have historically outperformed lower-income students on standardized exams, largely because of an imbalance in tutoring. According to The Washington Post, “Students from families earning more than $200,000 a year average a combined score of 1,714 [on the SAT], while students from families earning under $20,000 a year average a combined score of 1,326.”
AI-driven Privilege Expansion in education may close this gap. Everyone should get their own tutor in their pocket.
Healthcare Professionals (Low-Acuity)
Healthcare suffers from a similar problem to education: limited access to professionals. And when access is possible, we run into cost barriers. There’s a lot of demand competing for scarce supply.
AI expands access by being that supply—by rendering the human moot, particularly in low-acuity cases. Think of the categories mentioned in last spring’s The Telehealth Tipping Point, shown again here; each category is ripe for disruption from an AI healthcare professional.
Again, AI use cases won’t be the complex, high-stakes. But can a Nourish AI bot recommend a personalized meal plan? Can a Honeydew AI bot recommend the right spot treatment for acne? Can a Headway AI bot offer tips on how to deal with stress? Who knows if the winners will come from these companies, which today primarily offer access to a human via telehealth—but I imagine each will soon have an AI alternative to expensive, hard-to-reach AI professionals. The humans can focus instead on the more challenging, intuition-driven cases.
Dazed ran a headline this week, “Meet the People Using ChatGPT As Their Therapist.” The headline is a bit alarmist—“Shouldn’t people, you know, have access to a trained professional?,” you might be thinking. Well, yes, but that’s expensive and we run into the limiting variable of the human. AI can serve low-acuity cases in a more economical and efficient manner.
In healthcare, we’re going to see agents on the B2B side (helping doctors, nurses, hospital admins) and we’re also going to see consumer-facing AIs that meet many healthcare needs.
Stylists
In The Egg Theory, Applied earlier this month, we covered one of our stealth Daybreak companies that’s building a stylist for everyone, using AI. In some ways, this is the natural evolution of Stitch Fix, except the human stylist is now an LLM, which is more cost-effective and more scalable.
Instead of searching for “women’s white blouse” or “men’s chino pants in medium,” you might query, “I’m looking for outfits to wear to an art gallery opening in Bushwick” and voila, your AI stylist recommends products to you. In my mind, this is the biggest shift in commerce: new technology allowing a more natural form of discovery-driven shopping. Stylists for everyone.
This is another example of Privilege Expansion: a high-end service, previously cordoned off for the wealthy, now made accessible to the masses using technology.
Interior Designers
We see a similar theme for interior design—hiring a human interior designer isn’t exactly cheap.
The data in this chart comes from ANGI, and I imagine it’s not perfect—but it certainly makes the point, which is that access to an interior designer (as with access to a personal stylist) is limited only to the wealthy.
Now, everyone should get this privilege.
Travel Agents
We’re also seeing innovation in the travel space—“travel agent” is another human-centric service ripe for disruption. Companies like Mindtrip offer AI-powered trip planning. Mindtrip uses a chat interface, where you enter information about your trip and your preferences; Mindtrip then outputs personalized itineraries and recommendations.
High-Paid Knowledge Workers (Lawyers, Bankers, Accountants, Etc.)
July’s AI Is a Services Revolution argued that AI will reinvent knowledge work. America’s economy has shifted from an agriculture economy to a manufacturing economy to a services economy.
Services are dominating the 21st century workforce—check out this data from the Bureau of Labor Statistics:
The July piece dug into areas like law, investment banking, and insurance—how each will be reinvented by AI. These are expensive services, limited to the few. We should see Privilege Expansion here. As one example: I’m seeing a lot of innovation in Personal Injury law. This is a big subset—IBIS estimates the ‘Personal Injury Lawyers & Attorneys’ industry to be worth $57B a year in revenue. But in spite of all those highway billboards, it can still be hard to find an affordable personal injury attorney. I imagine AI can handle a good deal of consumer use cases here, dramatically lowering cost, wait time, and other barriers to access.
And…Friends?
There’s another service we don’t talk as much about: friendship.
AI is already seeping into how we form friendships and interact with friends. Daily actives spend two hours a day on Character, with average session length hovering between 25 and 45 minutes. Many other similar products have emerged to let us chat with AIs.
Or take an even more dramatic example: Friend.com, a company that went viral for its launch video a month ago—and for spending a huge portion of its Seed capital on purchasing the Friend.com domain.
Friend combines hardware—a pendant worn around the neck—with software to connect you to an always-on friend, an AI that can react to your day-to-day life, offer companionship, and, as shown in the video, occasionally roast you. Of course, Friend.com struck many people as something straight out of a Black Mirror episode.
Yet one can argue that friendship is, in some ways, a privilege. Not everyone has close social ties. Here’s a depressing chart, from a prior 10 Charts edition:
In the last 30 years, the percentage of people with <=1 close friend has nearly tripled to 20% of the population. The percentage of people who report 10+ close friends has dropped from 40% to 15%.
Is AI the right solution here? Probably not. It would be nice to, you know, give people real human companionship. But friendship is a service gated by access to a human—and it’s a service that AI will no doubt attempt to broaden access to, giving the friend-less among us the “privilege” of a friend, evening if that friend takes an artificial form.
Final Thoughts
There was a debate on Twitter last week after Y Combinator’s Michael Seibel tweeted this:
He’s right. Any consumer-facing service is ripe for disruption. I liked Sarah Tavel’s reply, which showed that we’re already seeing a dramatic shift in consumer behavior. Before AI, she pointed out, the fight for consumer minutes was next to impossible: BeReal, the most successful consumer social startup in recent memory, could barely get two minutes of engagement per daily active user. It’s tough to compete with TikTok, Instagram, Roblox, Discord, Candy Crush.
Now, though, we see consumers spending hours on Character and similar products, chatting with AI companions and friends. That’s a massive shift in attention.
By removing the human, we remove artificial barriers of time and money. This is the formula for Privilege Expansion: things many of us couldn’t previously get, we now get (1) immediately, and (2) for the fraction of the price. Return to the formula:
Expensive Human-Centric Service + AI = Better Access & Affordability
I expect dozens of generational companies to be born from this formula.
Sources & Additional Reading
Meet the People Using ChatGPT As Their Therapist | Dazed Digital
The Most Misread Poem in America | Paris Review
Thanks for reading! Subscribe here to receive Digital Native in your inbox each week: