Perlego’s Position on AI
Understand the principles guiding our use of AI tools and our commitment to content rights and transparency
How do we use AI?
We us AI in several ways across our platform:
- AI Curation:
- We have strict curation rules to ensure that only the best academic content ends up on Perlego. Until recently, this was done almost entirely by one Perlego employee. Thanks to AI, our curation is now automatic and much faster!
- Content Discovery and Recommendations:
- We partnered with Shaped.ai to build smarter content recommendations using real-time behavioural signals. Users will see titles suggested to them based off their previous reading, which leads to higher engagement and better discovery of relevant content.
- Our Research Assistant Tool:

- Our Research Assistant tool uses a RAG model to augment our search experience with AI, helping publishers earn more revenue and students discover verified content.
- When a user asks a question, we run a Deep Search to retrieve the most semantically relevant passages - small chunks of text within a book - for the given question.
- We then combines the user’s original question along with the semantically relevant passages within a prompt, similar to “answer the question below using only the information in the provided context”. The prompt is sent to an LLM run on our servers.
- The LLM will provide an answer grounded in the context provided (the semantically relevant book passages), separate each claim and attribute it to source it used. The referenced answer is finally returned to the user.
- Want to see Research Assistant in action? We recently hosted a webinar to showcase this innovative tool and its potential applications. Watch the recording here.
-
Our Read Aloud feature
- Perlego offers the ability for users to listen to their books as they read. We use AWS Polly’s neural voices to convert text to speech so that any book can be listened to, including both ePub and PDF formats.
Do we train AI on the book dataset?
Perlego does not use publisher content to train LLM models, nor is it sent to third-party AI provider companies like OpenAI or Anthropic.
Do we accept AI-generated content on our platform?
We do not currently have the ability on Perlego to flag that content is AI-generated. We want our users to be able to make an informed decision on whether they use AI-generated content for their studies. As such, we do not accept this type of content. If we decide to add flags that signpost AI-generated content in the future, then this is a position we may change. Until then, we kindly ask that you do not knowingly send us these titles!
Our guiding principles on AI
These 4 Principles are designed to ensure we embrace AI thoughtfully, responsibly and collaboratively with the education and publishing community:
💪 Support rather than replace: We implement AI to support and enhance, not replace, the learning experience.
💛 Collaboration & trust: We create trust by consulting our HigherEd & publishing partners on AI.
🙅 Mitigate against bias: We work hard to prevent the introduction of unfair bias through AI on Perlego.
✋ Privacy as the standard: We respect the privacy and intellectual property of our users and partners.