Rajiv Garg

Working paper · 2024

The Impact of Generative Artificial Intelligence on Higher Education: Disruption or Seamless Integration?

Students learned best from expert-written course content delivered by an AI voice and avatar. AI-written content scored lower no matter who delivered it.

Rajiv Garg

Read the paper

Key findings

  1. Human-written content with AI delivery produced the highest quiz scores (89.2%), ahead of a fully human course (84.9%).
  2. A fully human course outscored a fully AI-generated course by 5.7 points.
  3. AI-written content delivered by a human voice scored lowest (76.5%).
  4. The design separates who writes the material from who delivers it, across four versions of an online SQL course taken by about 70 undergraduates.

In plain English

Most debate about AI in classrooms lumps two questions together: can AI write good teaching material, and can AI deliver it? This study pulls them apart.

The same online SQL course was built four ways: human or AI content, delivered by a human or an AI voice and avatar. Students took one version and were then quizzed.

The answer split cleanly. Content quality depended on the human expert. AI-written material scored lower no matter who delivered it. Delivery did not need a human. An AI voice reading expert-written content beat the all-human version. One likely reason: many students watched at 1.5× speed, where a steady synthetic voice was easier to follow than natural speech.

Why it matters for leaders

  • For universities and corporate learning teams: Spend expert time on the curriculum, and consider AI for production and delivery. That is where AI saves cost without hurting outcomes.
  • For AI strategy generally: The value of generative AI depends on the task. A blanket “use AI” or “ban AI” policy misses the split between authoring and delivering.
  • For product teams building AI tutors: Learner outcomes, not engagement, should be the metric. Here the version that sounded most “human” was not the one that taught best.

In the press

Cite this paper

Rajiv Garg (2024). The Impact of Generative Artificial Intelligence on Higher Education: Disruption or Seamless Integration?. Working paper.

@unpublished{garg2024genai,
  title = {The Impact of Generative Artificial Intelligence on Higher Education: Disruption or Seamless Integration?},
  author = {Garg, Rajiv},
  note = {Working paper},
  year = {2024}
}