The purpose of the chatbot is to help you easily access and interact with the many reports, podcasts, blogs, videos, and other content created and hosted by Give.org about how to trust charities, and general advice on giving and philanthropy.

Users should understand that this is an early stage technology and there are going to be mistakes. We ask for your collaboration in helping us improve the chatbot by sharing your experience in the feedback slot.

Yes, Give.org stores all data from this chatbot permanently to understand how people are using this new technology, and how to improve the service.

No, the chatbot is not intended to make any direct suggestions about what charity to donate to. It's only meant to Give general information about charities, philanthropy, or giving based on the information that Give.org already publicly offers on its website.

No, the bot cannot answer questions quantitatively. For example, if you ask it something like, "How many charities in the Give.org database are there that have to do with animal rights?" it won't be able to answer this.

As the chatbot uses generative AI, it is possible for it to generate information on its own by accident. If you suspect that something is inaccurate or malformed, please use the "Report" feature on the chatbot and provide details on the issue. you should take a screenshot and report it to Jeremy at handshake.fyi.

No, the Give.org bot does not contain all of the information that can be found in the charity reports on Give.org. People are encouraged to also check out the Give.org charity reports directly after using the bot.

The Give.org chatbot is built on top of Anthropic's Claude 3 Sonnet model, along with custom tooling built by Handshake for Give.org.

No, Anthropic does not store any data permanently. They keep the data put into the bot for 28 days for debugging purposes and then delete it. It is never used to train any AI models.