by The $57k robot teacher: did physical AI just hit labor cost-parity?
A New York school district bought a humanoid named Sally for less than a teacher's salary. The internet turned it into a joke. Under the jokes is the question every physical-AI investor is trying to price: when a robot costs less than the human it stands in for, is that the milestone — or is the price tag the only real thing about it?
- Salamanca City Central (NY) is piloting "Sally," a $57,000 Realbotix humanoid, in 11th/12th-grade AI and robotics classes — priced explicitly below a local teacher's salary.
- The bull read: this is the first publicly-priced deployment of an embodied humanoid into a salaried role. Realbotix's CEO called it "moving beyond lab demonstrations" — cost-parity as the tipping point for labor substitution.
- The bear read: Sally can't walk, handles one student at a time, is a rebranded companion-robot chassis (Realbotix bought sex-doll maker RealDoll in 2024), and lands in a district that couldn't staff the class at all. A staffing gap, not a substitution.
- The TT desk call: price parity is not capability parity. The value here accrues to the AI teaching layer and to whoever solves mobility and dexterity — not to bolting a face onto a chatbot. Pass on the humanoid-shell narrative.
The facts under the noise
Strip out the sex-doll jokes and the culture war, and the reporting is specific. Salamanca City Central School District, on Seneca Nation land in western New York, has partnered with Toronto's Realbotix to put a seated humanoid — Sally, part of the company's M-Series — into AI, coding and robotics courses, paired with an AI assistant called Optio offering 24/7 homework help. Sally recognizes students by ID, remembers past sessions, moves her arms, and cannot walk around the room. The program runs under the Woz ED STEM curriculum started by Apple co-founder Steve Wozniak. Realbotix CEO Andrew Kiguel called the deployment "a landmark moment for both AI and humanoid robotics… moving beyond lab demonstrations and pilots to deliver real, embodied AI directly into classrooms."
one-time price of "Sally," set explicitly below a NY teacher's salary
Realbotix · MashableNY teacher salary — recurring, plus benefits, pension, union
per threadSalamanca students economically disadvantaged; a third Indigenous
NY Focus · Mashableyear Realbotix acquired sex-doll maker RealDoll and rebranded
MashableFigures from the Mashable report ↗ and the source thread. Sally is a pilot, not a signed teacher-replacement contract.
The chain of the argument
The seed was Collin Rugg's post laying out the price tag and the vendor's back-story, ending "What are we doing here?" From there the thread forked. One camp read the price as the point: a humanoid now costs less than the human, and cost, not sentiment, is what decides these things. The other camp read the details as the point: it cannot walk, it teaches one kid at a time, and it comes from a company that makes intimacy robots. @cyber_razz did the cold math out loud — $57k once versus $60–80k a year with no benefits, pension or union — and concluded "Sally is just the first one." @k12careers reframed the whole thing: the district isn't replacing a teacher, it can't find one. "'Costs less than a teacher' only reads as a flex when there's no teacher to hire."
The story isn't a robot replacing teachers — it's that districts can't staff CS and robotics at all. A staffing gap wearing a sci-fi headline.
The two sides
For — this is the cost-parity milestone
Realbotix & the substitution camp
"A landmark moment… moving beyond lab demonstrations and pilots to deliver real, embodied AI directly into classrooms." The vendor is selling this as the start of a new era, not a stunt.
A. Kiguel / RealbotixThe budget math is brutal and one-directional: $57k once, no salary, no benefits, no pension, no union. A person costs more every year; Sally doesn't. "Sally is just the first one."
@cyber_razz"Sally costs less than the average teacher" and "robots will never replace teachers" can't both hold — the drive to save money will override the promise.
@KiwiThinkerTeaching jobs get scarce from here — AI does the instruction and grading first, and robotics catches up to the physical role after.
@SellTowerAgainst — it's price theater, not embodiment
The skeptics
This isn't replacement, it's a staffing gap. Districts can't hire for CS and robotics; "costs less than a teacher" only reads as a flex when there's no teacher to hire.
@k12careersThey took an LLM tutor — which scales to every student at once — and bolted it into a body that can only address one student at a time. The humanoid form is a downgrade.
@MikeDKirbySally is a Realbotix chassis — the same firm that bought RealDoll. This is a rebranded companion robot, not purpose-built ed-tech.
@TRIGGERHAPPYV1The demo's own pitch is "I make emotional connections" and "how not to be lonely with AI" — companionship marketing, not pedagogy.
@kalinndii_If the job is to teach, the anthropomorphic shell adds nothing. Build it like a box; the human face is cost and distraction, not capability.
@OannesInNinevehWhy this is an investment question, not a meme
The humanoid trade rests on one number: the point where a robot costs less than the worker it replaces. Every deck about physical AI draws that crossover and dates the boom to it. Sally is the first time that crossover shows up with a real invoice attached to a real institution — which is exactly why it's worth reading carefully instead of laughing past it. The problem is that "cheaper than a teacher" answers a price question and dodges the capability one. A thing that costs less than a worker and also can't do the worker's job isn't a substitute; it's a subsidy line item. The skeptics aren't being snobs — they're pointing at the two failure modes that break the whole thesis: the demand is a shortage (nobody to hire), not a displacement, and the embodiment is a regression (one student at a time) versus the software it wraps.
The TT desk thoughts
Price parity is not capability parity, and the market keeps conflating the two. Sally clears the price test and fails the capability test — she can't walk, can't manage a room, and does one-to-one what an LLM does one-to-many. The investable signal in this story is not "humanoids replace teachers." It's two quieter things. First, the pull is a labor shortage, which is bullish for the software layer — the Optio-style AI tutor that scales to every seat — and roughly neutral for the $57k chassis it rode in on. Second, the moment a humanoid body is genuinely worth its cost is the moment it does something a screen can't: move, manipulate, work unsupervised in physical space. That's a mobility-and-dexterity problem, and it's where real physical-AI value will accrue — the same reason hands, not faces, are the bottleneck. So: fade the anthropomorphic-shell narrative when it's priced as the milestone. Own the AI teaching/agent layer and the companies solving locomotion and manipulation. A seated robot with lip fillers filling a hiring hole is a headline, not a comp. When the humanoid can walk the floor and do the job unattended, re-underwrite it — until then, the face is marketing and the software is the asset.
Keep reading
Is physical AI investable? — the framework · 1X Neo and why hands are the bottleneck · All notes