The first problem with an algorithm is often the room in which it was discussed. Inside are the people designing it, testing it, perhaps selling it. Outside are the people whose grades, work or creative choices may be shaped by it. Jordan Harrod has spent eight years walking between those rooms with a camera, a scientist’s habits and a talent for asking questions that sound simple until someone has to answer them.
Her YouTube channel began in the summer of 2018, after she graduated from Cornell. She wanted to explain artificial intelligence to people who were absent when decisions about it were made. The channel was a practical answer to a practical gap. She had looked for casual, accessible explanations of the algorithms people encounter every day and found too few. So she made her own.
There is an appealing modesty to that origin. No founding myth, no claim to have seen the future. A young engineer looked at a subject wrapped in jargon and thought that more people deserved a way in. By 2026, the channel counted more than 89,000 subscribers, 2.4 million views and over 250 videos. Numbers can describe an audience; they cannot quite describe the relief of finding someone who will explain what a system does before telling you what to think about it.
A lesson before the lecture
Harrod studied engineering at Cornell, where she also spent time working with machine learning. At Stanford during an undergraduate summer, she learned parts of that field as she went. She later entered the Harvard–MIT Health Sciences and Technology program, drawn to work that could travel beyond a single discipline. The doctorate took eight years. Her channel lived alongside it, rather than waiting politely for the diploma.
That overlap explains the texture of her videos. She speaks with the confidence of someone who has had to test an idea, and with the caution of someone who knows how easily a headline can outrun its evidence. In 2021, she described the need for people to see a fuller picture before making decisions. For a video maker, that is a demanding promise. A neat answer is easy to clip. A faithful answer may require explaining why a result is narrow, what it leaves out and why the viewer should still care.
Her 2019 TEDxBeaconStreet talk made a larger argument: understanding AI would become part of ordinary technological literacy. At the time, a conversation about AI could still sound like a conversation about distant laboratories. Harrod was already pointing toward the systems people would meet in their schools, feeds and offices. Literacy, in her telling, meant recognizing where the technology was operating and having enough language to ask what it might change.
When a classroom became a test case
In May 2020, she published a video about automated online exam proctoring. The subject had suddenly acquired an enormous audience: remote learning was making software a watcher in students’ rooms. The title asked a blunt question: could AI proctors detect online exam cheating? It was a sharp choice of subject for someone interested in the human side of technical decisions. Students could be judged by a system they might know almost nothing about, at a moment when the rules of school had already been rewritten around them.
The video has since drawn hundreds of thousands of views. Its enduring interest lies beyond the particular products. It is an example of the kind of encounter Harrod notices: a technical system moves from a developer’s demonstration into an ordinary person’s consequential day. The public then has to catch up with the product. Her work tries to shorten that distance.
A very different lesson arrived through a collaboration with Tom Scott. He challenged Harrod to make a synthetic version of him on a budget of $100. The result was playful and faintly unnerving, which made it useful. A deepfake stopped being a vague threat in an article and became a production problem with a price tag. Harrod used the exercise to talk about how such media is made and what its accessibility could mean. The joke got the audience in; the mechanics made them stay.
“They deserve to know the full picture so they can make informed decisions.”Jordan Harrod
She also moved some of her work to Nebula, the creator-owned streaming platform she now co-owns. She has noted that covering subjects such as automated moderation could itself run into moderation systems. There is a tidy irony there: a creator explaining the rules of an algorithm may have to negotiate those rules to reach an audience. It is less tidy when the creator’s work depends on it.
The questions she takes into the room
Over time, the audience for Harrod’s explanations widened beyond viewers. Her consulting practice, Harrod & Co, has worked with organizations ranging from YouTube/Google and the Gates Foundation to smaller businesses around Boston. That work included a 20-month residency with YouTube/Google. The work includes looking at AI decisions before launch and helping institutions communicate with audiences that have reason to be skeptical.
This is where the channel and the advisory work meet. Someone who has spent years reading public reactions to AI systems brings a different set of questions into a product conversation. Harrod has described asking what a proposed feature is meant to change, why creators say they want it and how it could go wrong. These are not dazzling questions. Their value is that a product team must answer them in plain language, with actual people in mind.
She describes the same project from the other side: “I help people and institutions make sense of AI.” The sentence has the unusual virtue of fitting both a video watched alone and a meeting with a large organization. In one setting, she gives someone the vocabulary to judge a tool. In the other, she asks the makers to explain their judgment before the tool reaches anyone else.
Harrod’s writing and appearances have carried this argument into other formats. She contributed an essay on algorithmic fairness to The Black Agenda in 2022. She has given talks, appeared in interviews and written a newsletter. These formats do different things. A short video can catch the person who has never searched for an explanation. An essay can follow the consequences further. A course can leave room for someone to say, without embarrassment, that they do not understand a term everyone else seems to know.
A graduation, then another question
In the summer of 2026, Harrod announced that she had finished her Harvard–MIT PhD. The announcement had a little comedy of academic life in it. She joked about the price of ceremonial regalia and how little use the hat and hood would likely get afterward. Eight years of work ended with a degree, a certificate and a video maker’s instinct to explain the odd details to the people who had followed along.
That continuing relationship with viewers is part of the story. She started the channel just before graduate school, and many subscribers watched both projects unfold together. She has since built AI IRL, a discussion-based course for people who want a framework for living with AI. She also runs workshops and is developing a book, How It Actually Works, about governing AI through democratic institutions. The planned book points toward a larger version of the question she has asked since 2018: who understands a decision well enough to take part in it?
She remains connected to Cornell through the College of Engineering Alumni Association board. Nebula gives her another stake in how educational creators reach audiences. The consulting practice brings her into institutions where AI ideas may still be drafts rather than finished products. None of these roles makes the others redundant. Each offers a different moment to intervene: when an idea is forming, when a system is being explained, or when a person is trying to decide what to do with it.
There is a temptation to call anyone who explains AI a translator. It is true as far as it goes, but it can make the work sound too passive. Harrod’s career suggests a more active job. The translator can ask why a sentence was written, whom it was written for and whether it should be sent at all. As AI products move into ordinary routines, those questions become as useful as a glossary.
At the end of one video, the viewer may know how a proctoring system works. At the end of a meeting, a product team may know that a feature needs a clearer purpose. Between those moments is the room Harrod has been trying to open since her first upload. The door is ordinary enough: a person asks a question and someone takes the time to answer it properly.