From Mainframes to Mobile

How learning technology was born — and how it changed everything

Juliette Denny

Founder, Zavmo AI

7 Apr 2026

10 min read Beginner

Is This For You?

What You'll Learn

Key Topics

In 1958, Burrhus Frederic Skinner published his groundbreaking work on programmed instruction, introducing a revolutionary concept: that learning could be systematically designed, delivered, and measured through machines. More than sixty years later, that simple idea has transformed into a global industry worth billions, reshaping how organisations invest in their people. Yet the journey from Skinner’s paper tapes to today’s mobile-first learning platforms reveals far more than technological progress—it exposes the fundamental tensions between content delivery and human development that continue to shape the field.

The Birth of Computer-Based Learning (1960s)

The earliest attempts to use computers for education emerged not in universities, but in the military and aerospace industries. In the early 1960s, researchers at the University of Illinois under the direction of Donald Bitzer and his team developed PLATO (Programmed Logic for Automated Teaching Operations), a system that is widely recognised as the first networked computer-based education system. Beginning its operational deployment in 1960, PLATO was revolutionary: it combined Skinner’s principles of programmed instruction with networked terminals, allowing multiple learners to access content simultaneously and enabling instructors to track individual learner progress in real time.

What made PLATO remarkable was not merely its technological sophistication, but its pedagogical integration. The system featured branching logic—learners could follow different learning paths based on their responses—and embedded assessment. PLATO established a design pattern that would persist for six decades: content delivery, learner interaction, and performance data collection as an integrated system. The aviation industry, facing acute training challenges in the 1960s and 1970s, became an early adopter of PLATO and similar computer-based instruction systems, recognising that simulation and standardised measurement offered cost and safety advantages over traditional classroom methods.

The Expansion of Computer-Based Training (1970s–1980s)

Throughout the 1970s and 1980s, Computer-Based Training (CBT) evolved from an experimental curiosity to a commercial industry segment. The shift from mainframe terminals to personal computers and, critically, the emergence of CD-ROM technology, made delivery practical at scale. Organisations could now produce self-contained training modules that did not require networked infrastructure—a massive advantage during a period when many companies lacked the technical capability to maintain internal networks.

In 1988, the aviation training community formalised its standardisation efforts through the formation of the Aviation Industry CBT Committee (AICC). This organisation would become foundational to the industry’s technical infrastructure. The AICC developed guidelines for content packaging, runtime communication protocols, and learner tracking—establishing the technical vocabulary that would eventually underpin all subsequent learning technology standardisation efforts. What the AICC recognised, which many in the broader learning technology community had not yet articulated, was this: standardisation creates markets. If content could be built once and deployed across multiple platforms from multiple vendors, an entire supply chain would emerge.

The Web Revolution (1990s)

The emergence of the World Wide Web in the 1990s fundamentally restructured the learning technology industry. For the first time, there existed a universal, platform-agnostic delivery mechanism that did not require proprietary terminals or expensive infrastructure investment. The shift from CD-ROM to browser-based delivery was not merely technological; it was organisational. Training departments could now deploy learning solutions to geographically dispersed workforces without requiring every learner to have compatible hardware or software installed locally.

The first Learning Management Systems (LMS) platforms emerged during this period—systems designed to manage the entire lifecycle of learning delivery: enrolment, content hosting, learner access, tracking, and reporting. Systems such as Blackboard (founded 1997) and WebCT initially targeted higher education but quickly found adoption in corporate training environments. These platforms solved a critical business problem: how to scale learning delivery across a growing organisation without proportionally scaling training personnel.

Yet even as these platforms proliferated, a critical problem emerged. Each vendor used proprietary data formats and content packaging standards. A course built for one LMS could not easily be deployed on another. This created vendor lock-in: organisations investing in content development became dependent on their chosen platform, and switching platforms meant abandoning existing investments or investing substantially in content migration and reformatting.

Standardisation and the SCORM Era (Early 2000s)

In 1997, the United States Department of Defense initiated the Advanced Distributed Learning (ADL) initiative, recognising that the military faced a critical challenge: maintaining training standardisation and interoperability across thousands of training providers and millions of personnel distributed globally. The ADL initiative, funded by DoD but conducted as an industry collaboration, set out to develop technical standards that would enable content interoperability.

The result was SCORM (Sharable Content Object Reference Model), version 1.0 released in January 2000. SCORM represented a watershed moment in learning technology. It standardised three critical elements: content packaging (building on the IMS Content Packaging specification), runtime communication (establishing a standard data model based on IEEE 1484.11.1), and learner assessment data. For the first time, organisations could theoretically purchase content from one vendor and deploy it on an LMS from another vendor, with learner completion and scoring data travelling reliably between systems.

SCORM’s impact was immediate and profound. By 2002, it had achieved near-universal adoption in regulated industries where compliance training was mandatory—aviation, healthcare, financial services, and defence contracting. The military, in particular, drove widespread adoption. The Department of Defense’s decision to mandate SCORM compliance for all its learning content suppliers created a cascading requirement throughout defence contracting and aerospace industries. If a company wanted to bid on military contracts, it needed SCORM-compliant content and platforms.

Denny (2025) argues that the standardisation movement in learning technology was driven fundamentally by economics rather than pedagogy: the goal was not to improve learning but to reduce costs and increase efficiency through content reusability and vendor competition on features rather than format. This distinction—between learning outcomes and operational efficiency—would prove critical to understanding both SCORM’s success and its eventual limitations.

Early Adoption and Market Drivers

The rapid adoption of SCORM and Learning Management Systems in the early 2000s was driven by four converging forces. First, regulatory mandates created compliance training requirements that organisations were legally obligated to document and evidence. The financial services industry, facing securities regulations that required documented compliance training, invested heavily in LMS infrastructure. Healthcare organisations needed to track continuing professional development and competence assessments. Aviation required standardised training documentation and currency records.

Second, the cost advantage of technology-enabled training compared to classroom delivery became mathematically compelling. A single instructor could reach thousands of learners through an online platform; classroom training required proportional increases in instructor headcount. For organisations with geographically distributed workforces, the travel costs alone made classroom training prohibitively expensive compared to online alternatives.

Third, the globalisation of large organisations created a need for standardised, globally distributed training. A multinational corporation with operations across fifty countries could not practically deliver consistent classroom training; online platforms with translation and localisation capabilities provided a solution.

Fourth, early research on learning outcomes suggested that properly designed technology-mediated instruction could match or exceed classroom instruction effectiveness. Clark’s media comparison research in the early 1980s had been controversial, arguing that media itself was neutral—that learning outcomes were determined by instructional design, not delivery medium. However, Bloom’s research on tutoring effects published in 1984 provided compelling evidence that intensive, personalised, adaptive instruction could produce learning outcomes approximately two standard deviations above conventional classroom instruction. This finding animated industry research throughout the 1990s and 2000s: could technology deliver tutoring-like personalisation at scale?

The Academic Foundation

The early research base on learning technology was sparse but influential. Benjamin Bloom’s meta-analysis demonstrating that one-to-one tutoring produces approximately 2 standard deviations of improvement over conventional classroom instruction became a north star for the industry. If technology could deliver tutoring-like adaptation, Bloom’s findings suggested the potential was transformative.

Clark’s research on media comparison, whilst controversial, shaped how educators thought about technology. Clark argued that media selection should be driven by practical and economic considerations rather than assumptions that particular media types would produce inherently superior learning. This perspective, whether fully accepted or not, became foundational to how organisations evaluated learning technology investments: the question was not whether technology was superior to classroom instruction in some absolute sense, but whether it was more efficient and effective for particular organisational contexts and learner populations.

During the 1990s and 2000s, instructional design theory became increasingly sophisticated. Merrill’s component display theory, Clark and Mayer’s multimedia principles, and Sweller’s cognitive load theory provided frameworks for designing effective computer-based learning experiences. The field was developing not just technologies but pedagogical principles for using those technologies effectively.

Market Growth and Maturation

The learning technology market grew rapidly through the 2000s. Early estimates from the mid-1990s valued the global training market at a few million dollars. By 2005, industry analysts estimated the corporate e-learning market at over USD 20 billion globally, and the figure continued to climb. Major software vendors—SAP, Oracle, IBM—integrated learning management capabilities into their broader human capital management platforms. Dedicated learning technology vendors raised significant venture capital and achieved multibillion-dollar valuations.

Yet even as the market expanded, critical questions about learning effectiveness remained unanswered. Did online training actually improve job performance? Were learners in virtual environments learning as effectively as learners in classrooms? Were organisations measuring learning at all, or merely measuring completion?

The Structural Foundation

What the learning technology revolution of the 1960s through 2000s established was a structural separation of concerns that would persist for two decades. On one side stood learning technology infrastructure: systems designed to manage content, track learners, and generate compliance reports. On the other side stood learning content: the actual educational material, whether video, interactive simulations, assessments, or text-based modules.

This separation made sense from engineering and business perspectives. Large organisations could invest in LMS infrastructure once and leverage it across many business units. Content providers could develop courses that would work on any compliant LMS. Vendors could specialise: some built platforms, others built content, others built integrations.

Yet this structural separation embedded an assumption that would later prove problematic: that learning was fundamentally a content delivery problem. If organisations could simply get the right content in front of the right learner at the right time, learning would follow. This assumption worked reasonably well for compliance training, where the goal was to document that learners had been exposed to required information. It worked less well for knowledge and skill development, where the processes of learning were far more complex than content consumption.

Conclusion: The First Generation

By the mid-2000s, the learning technology industry had matured through its first generation. It had developed standardised platforms, established technical interoperability standards, created a robust supply chain of content and platform vendors, and achieved mainstream adoption in large organisations worldwide. The early promise of computer-based instruction—that learning could be systematically designed, efficiently delivered, and reliably measured—had been substantially realised.

Yet as Denny (2025) observes, the industry had achieved operational efficiency without necessarily advancing learning science. Organisations had invested billions in infrastructure that allowed them to deliver training at scale and measure completion and compliance. Whether learners were actually developing knowledge and skills that transferred to job performance remained, for most organisations, unmeasured and unknown. The next generation of learning technology would attempt to address this gap, asking not merely how to deliver content, but how to fundamentally transform how people learn.

Stop the Forgetting Curve

See how Zavmo’s AI-powered spaced repetition and retrieval practice make learning stick — for professionals and students alike.