A first that has several claimants
The title assumes there was one first computer-based education system. In practice, several university and company projects began in the late 1950s and early 1960s, and they influenced one another. Which one counts as first depends on definitions: the first lesson program, the first system serving many students, or the first used regularly in real schools. For that reason this article describes the group of efforts and avoids naming a single winner.
What they shared is the approach now called computer-assisted instruction: a computer presents material, asks questions, checks answers and decides what to show next.
The ideas behind it
Two earlier threads fed into the work. The first was programmed instruction, popular in the 1950s, in which a subject is cut into small steps and students respond to each one. Mechanical “teaching machines” and printed programmed texts already used this idea. The second was the growth of university computing, where researchers wanted uses for machines beyond calculation.
A computer seemed to improve on printed programmed texts because it could branch. A student who answered correctly could skip ahead, and one who made mistakes could be sent to review. In theory this allowed individual pacing for large numbers of learners, which a single teacher with a full class could not easily offer.
The main projects
Exact histories vary by source, so only the broad outlines are given here.
University of Illinois. Work on PLATO began around 1960 and developed through several versions over the following years. Later versions connected many terminals to a central computer and used a distinctive plasma display. A programming language called TUTOR was developed so that teachers and subject experts, not only computer scientists, could write lessons. PLATO also became known for features beyond lessons, such as messaging and games, though those belong to a later chapter.
Stanford. Researchers there, working with a computer maker, developed drill-and-practice programs in subjects such as arithmetic and reading, and tested them with elementary school students in the 1960s. The emphasis was on supplementing a teacher’s classroom work with short, regular sessions at a terminal.
Company and other efforts. Computer makers and research labs built experimental tutoring programs of their own, and other universities tried similar courses. Some work was funded by government and private foundations. Because these projects used different machines and different lesson formats, they were not easy to compare or share.
What a lesson looked like
Early lessons were usually text-based. A student sat at a terminal, read a prompt, and typed an answer. The program compared the response to expected answers and replied with a message or the next question. Drill programs repeated arithmetic or vocabulary exercises, while tutorial programs explained a concept and then tested it. Graphics were limited, and the cost of keeping a central computer running meant that each terminal session had to be justified.
Authoring was a real obstacle. Writing a good lesson took a long time, and it was often unclear whether the result was better than a good textbook or teacher. This is one reason that authoring languages and authoring tools received so much attention.
Limits and criticisms
Computers of the 1960s were large and costly, so these systems reached relatively few students compared with the whole school population. Evaluation was also difficult. Studies of early computer-based instruction often reported modest benefits for drill-type material, but results depended on the subject, the students and the way the system was used, and it would be unwise to read much more into them.
Critics also questioned the model itself. If a computer simply asks questions and checks answers, it risks reinforcing memorization over understanding. That concern helped inspire a different approach, in which learners program computers themselves, as in the later Logo programming tradition.
Why it still matters
These projects set patterns that persist in educational software: adaptive questioning, immediate feedback, tracking of student progress and the separation of lesson authors from programmers. The hardware of the time seems primitive now, since these were mainframe computers shared among many users, but the questions they raised, such as what a machine can teach well and what it cannot, are still current.
Today’s online learning platforms draw on the same basic idea, even though they run over the internet rather than on a university’s private terminal network. The 1960s experiments were less a single invention than a decade of overlapping trials, and the credit is best shared among them.