The World History Encyclopedia and AI
I received an email this morning from the good folks at the WHE entitled: Perplexity AI Partnership to Improve History Education
In it, their CEO Jan van der Crabben, writes:
As the most-visited history encyclopedia globally, World History Encyclopedia is pleased to announce a strategic partnership with Perplexity.ai.
…
As artificial intelligence (AI) tools based on large language models become increasingly accessible to the public, growing concerns have emerged regarding the quality of information provided by these tools. These AI systems are typically developed and trained using publicly available internet information, often without robust verification processes, and frequently generate inaccurate results.
There are also significant concerns about the business models of AI companies, which utilise content developed and meticulously checked by providers like World History Encyclopedia —a non-profit organisation— without obtaining proper consent, without providing compensation, and without offering appropriate attribution.
Perplexity.ai is an AI-powered search and answer engine that combines the capabilities of a search engine with artificial intelligence. Unlike most other AI systems, Perplexity clearly cites its sources, providing users with an easy way to verify the accuracy of its answers.
In alignment with our goal of being a trusted resource of accurate and objective historical information, we are excited about this partnership. It will allow us to develop tools based on the Perplexity API to make the content in World History Encyclopedia easier to find, browse, and access. We aim to develop educational AI tools for history learning in close collaboration with teachers to augment the World History Encyclopedia website for students.
The partnership will also enable World History Encyclopedia to use artificial intelligence to enhance our human review processes more efficiently. This includes improving tasks such as fact-checking and plagiarism detection.
Additionally, Perplexity is the first AI service that allows providers of information like World History Encyclopedia to be compensated fairly for the AI use of that information. We will receive a share of advertising revenue generated on the Perplexity platform whenever Perplexity cites World History Encyclopedia to answer a question.
I have worked with Jan and his staff many times over the last six years and I find them eminently trustworthy and dedicated to education.
What does everyone think of this kind of partnership moving forward? I understand Perplexity might have a slightly different approach that certain folks find promising.
And what kind of content do we think this might be able to generate? I look forward to your comments.
In my experience, fact-checking and plagiarism avoidance are exactly the things that AI is not good at.
I can't say for certain, but the wording of this statement carries a strong undercurrent of "they're going to scrape our content anyway so at least this way we can get something out of it" to me.
I don't think fact-checking or plagiarism detection are use cases where AI like this is super reliable, but if by "enhance our human review process" they just mean "our human reviewers will have access to these tools when they're reviewing," I wouldn't be super worried. It's unlikely to be worse than existing plagiarism detection software, and the ultimate results are going to heavily depend on the skill of the human fact-checker with or without AI tools available to them.