Voshi Help Center
FAQs
A: While the answer depends on the subject, your writing skills, and various other factors, it may not be the right question. The Voshi platform is designed to allow the author and the AI to work together, producing content better than either of them could on their own.
A: It’s free! You can create your textbook and course materials free of charge through the Voshi platform. See our Pricing page for details.
Tutorial
Creating a textbook outline may seem daunting. Follow the input box to get the best results.
Creating a topic outline may seem daunting. Follow the input box to get the best results.
Creating a first draft may seem daunting. Follow the input box to get the best results.
Concerns
How can I be sure that ChatGPT won’t hallucinate incorrect information?
Yes, current large language Models (LLMs), including ChatGPT, sometimes hallucinate. That means they occasionally write statements that are factually incorrect or don’t make sense. Hallucinations happen because LLMs rely on probabilistic next-word predictions to generate text.
Voshi uses ChatGPT to power its text generation. While our prompt engine attempts to minimize hallucinations, they are still possible. As an author, it is your responsibility to fact-check the material before publishing.
Ideas present in many published works are likely to be accurate: commonly misunderstood concepts and unusual cases have a greater chance of containing inaccuracies and should be checked more thoroughly.
Does ChatGPT reproduce copyrighted materials? What protections are available for copyright infringement concerns?
Since LLMs are trained on vast datasets and generate content through probabilistic next-word prediction, their output will generally be original text—a probabilistic combination of many sources to generate something original.
Some, including the New York Times, have claimed that ChatGPT will sometimes regurgitate their copyrighted material verbatim. OpenAI, the makers of ChatGPT, say this only occurs through “intentionally manipulated prompts” by deliberately instructing the LLM to regurgitate the material or including large excerpts of the article in question. As Voshi’s prompts rely on your input as an author and do not intentionally seek to reproduce copyrighted material, it is extremely unlikely to do so.
“OpenAI is committed to protecting our customers with built-in copyright safeguards in our systems.”
If you have further concerns about copyright, feel free to reach out to us at concerns@voshi.com