Sally Isn’t the Problem. Sally’s Grandchildren Are.
A school district in Salamanca, New York — population 6,000, on the Allegany Indian Reservation — nearly spent $58,000 on a silicone-skinned android named Sally to help teach high schoolers this fall. Students helped design her: fair skin, long brown hair, youthful build. The manufacturer, Realbotix, has a corporate sibling in the sex doll business, which it acquired for $16.7 million.
Let’s take a moment to analyze and digest that statement. A school funded an AI robotic company with ties to ‘artificial girlfriends’:
Realbotix, formerly Tokens.com, a Toronto-based company, purchased Simulacra, a Las Vegas-based company known for realistic sex dolls, for $16.7m in April 2024, marking a pivot for the company from crypto to humanoid robots. [source]
There’s a lot of punchlines, possibilities , and lawsuits hinging on that quote but let’s move on.
Teaching robots are coming. Sally just showed up a little too early.
There are bad ideas. There are spectacularly bad ideas. And then there is spending nearly $58,000 to put a silicone-skinned humanoid named Sally in a high school and being surprised when people have questions.
That is essentially what happened in Salamanca, New York. According to Gothamist, Sally was supposed to arrive at Salamanca High School this fall as an AI-powered teaching assistant. She could hold conversations, make facial expressions and help students interact with emerging technology.
Then everyone noticed the elephant in the classroom. Or more accurately, the eerily lifelike brunette robot in the classroom. OK, maybe dirty blonde but those eyebrows tell us the real story.
There were questions about student privacy. Teachers worried about where this was heading. State education officials started asking uncomfortable questions. Then came the detail destined to make every school board meeting substantially more interesting: Realbotix, Sally’s manufacturer, has a corporate connection to a company involved in the adult-doll business.
This is generally not something you want to discover halfway through the PowerPoint presentation. The project was paused. And honestly, probably for good reason.
Sally represents almost everything wrong with our first instinct when approaching artificial intelligence. We take an interesting technology, slap a human face on it and immediately start looking for a job we can make it do.
But beneath the creepy silicone skin and the obvious jokes is a much more uncomfortable reality.
They are coming.
Maybe not Sally. Sally might spend the next few years packed in a shipping crate somewhere next to a broken Smart Board and 400 unused Chromebook chargers. But some descendant of Sally is going to walk into a classroom eventually.
And it might be pretty damn useful.
Imagine an AI teaching assistant 10 or 15 years from now that speaks 50 languages. It never gets tired of explaining fractions. It can spend forty minutes helping one student understand algebra while the human teacher works with the rest of the class. It could recognize that Johnny still doesn’t understand long division even though Johnny has become extraordinarily talented at pretending that he does.
A good AI tutor could adapt lessons to each student. It could help a kid at 9:30 at night when tomorrow’s homework suddenly looks like ancient Sumerian. In rural schools where advanced classes or specialized instructors simply aren’t available, that technology could be enormously valuable.
But here is the important part.
None of that requires pretending the robot is a person.
Teaching is a peculiar profession because transferring information is only half the job. Sometimes it isn’t even half.
A great teacher knows when a student is confused before the student raises a hand. They know when somebody is unusually quiet. They recognize embarrassment, boredom, fear, curiosity and the unmistakable expression of a teenager who has absolutely no idea what is happening but would sooner perish than admit it in front of his friends.
That’s not data transfer. That’s being human.
Artificial intelligence is getting frighteningly good at answering questions. Robots are getting better at movement, speech and facial expressions. Combine those technologies and eventually somebody will build a genuinely useful classroom machine.
But we’re not there yet. Right now, we’re in the awkward stage.
This is the educational equivalent of those first automobiles rattling down dirt roads while frightened horses stared at them. The technology works just well enough that you can see the future and badly enough that the present occasionally looks ridiculous.
The answer isn’t banning it forever. It’s also not dragging a $58,000 android into chemistry class because somebody attended a technology conference and got excited. Let teachers experiment with AI. Let students learn how it works. Build tutoring systems. Test robots in controlled environments. Establish privacy rules before collecting children’s data instead of holding a meeting afterward and wondering where everything went sideways.
And most importantly, keep the teacher.
For now, the sensible classroom of the future isn’t 25 children staring obediently at Professor RoboCop.
It’s a good teacher with extraordinarily powerful tools.
Twenty years from now, maybe the robot is standing beside them.
Fifty years from now, who knows? Maybe my idea for the Kotterbot 300 can gain some traction.
But when that day comes, somewhere in New York an old retired teacher is going to see the news, take a sip of coffee and mutter:
“Told you. It started with Sally.”