DeepHorus is an innovative tech media and intelligence platform focused on AI, blockchain, startups, and emerging technologies.
Published on

Academa – Long-form STEM lecture videos generated by LLMs

Authors
  • avatar
    Name
    DeepHorus
    Twitter

A new startup, Academa, has launched a platform that generates long-form STEM lecture videos using large language models (LLMs), aiming to revolutionize online education by treating videos as editable "source code." The core idea addresses a major limitation of traditional online education, where pre-recorded videos are static and difficult to correct once published. By converting the actions of a lecturer—speaking, writing equations, and drawing diagrams—into a textual, code-like format, Academa creates videos that can be compiled, edited, and improved over time, much like software.

The concept builds on the observation that the dominant format for online STEM education is video, citing platforms like Khan Academy and Coursera. However, producing and updating these videos is notoriously difficult and costly. Academa's approach transforms this process by first defining lecture content as a series of code-defined actions, which are then rendered into a video with text-to-speech and computer graphics. This maintainability is the foundational innovation.

The startup leverages the fact that LLMs excel at processing and generating text. By representing lectures as text-based code, Academa enables AI to directly create and manipulate the content. This unlocks several powerful possibilities. The technology can generate lectures on virtually any technical subject, from niche engineering topics to recent research, overcoming the scarcity of human instructors for specialized knowledge.

Furthermore, a single lecture coded in this format can be automatically translated into over 80 languages, producing first-class versions rather than simple dubs or subtitles. The system also paves the way for an interactive, AI-native learning experience. Students could pause a generated lecture, ask questions, and receive real-time video answers tailored to their understanding.

Academa has built the core lectures-as-code technology and is demonstrating it with LLM-generated videos. The startup states that these initial videos are just the beginning and will be maintained and improved iteratively. The company is actively developing new features and plans to share more details in the coming weeks and months.

While the full potential is yet to be realized, the initiative represents a significant conceptual shift in digital education. By merging the flexibility of software with the instructional power of video lectures, Academa is exploring a future where educational content is not just consumed but is also continuously updated, personalized, and globally accessible. The startup invites interested users to follow its progress through a newsletter and Discord community.

Read original article →Source: hn-frontpage