Each day we receive an overwhelming deluge of emails, direct messages and other reachouts asking questions about GDELT. While we try to help as many people as we can, there are simply too many inquiries for us to respond to everyone. Over the years we've found that around 95%+ of these emails are standard questions that could be answered by a simple Google search, by reading this blog or the documentation or by simply glancing at the response from the API. As but one example, we get at least a dozen emails on a typical day asking if we have a blog and suggesting that a blog could be useful.
Yesterday we got a 6-page AI-generated email that not only asked if we had a blog, but then proceeded to not only write 6 pages about what a blog is and why blogs are useful, but went on to explain what WordPress is, how to install WordPress on a server and configure it, and then, most amazingly, proceeded to link to recent blog posts on this very blog to describe how they could then be imported into a new "blog.gdeltproject.org" blog using WordPress to act as a blog.
This caught our eye because over the past year we've noticed that the percentage of daily emails we get that are clearly AI-generated is skyrocketing, with the majority clearly having not even been proofed by a human being. If we get a short email that explains an interesting problem and its societal significance and requests advice on how to best accomplish that goal using our data, we're only too happy to try to help. But when we get a machine-generated email the length of a small book that includes a multi-page introduction explaining what GDELT is and recommending that we "carefully consult the GDELT blog and related documentation to acquaint yourself with GDELT and what it does before responding to this email" (an actual line from a recent email), we are unlikely to respond.
While we appreciate the power of modern AI agentic assistants, please use them to research and answer your questions about GDELT, rather than have them simply draft an email to us. In the meantime, while we work towards launching an MCP server and other services to help AI agents navigate our datasets and APIs, a simple Google search or AI agent search can typically answer the majority of your questions.