Sure. So, with respect to AI, think we look at it in two ways. One is what you described, which is generative AI. How do you use AI technologies to assist in the creation of content? And when you think about the editorial workflow, there are a lot of elements to it, from origination of story idea, to assignment of story, to outline of story, to drafting of story, to copy editing of story. I certainly believe that artificial intelligence has a role in parts of that editorial workflow, and we've been using it and will continue to use it. I will say though, that I think for high quality content generation, human plus artificial intelligence, I think, yields the best outcome. So, I think the answer to your question is that generative AI will allow us to produce more better content over the long-term. It's a productivity gain more than anything else, which we're excited about and we currently use and have been using. The second part of AI, which is as, if not more exciting from my point of view for our company, is how you can take AI models, so right, ChatGPT is an AI model, and how do you then train it on proprietary data? The experiences we're all seeing right now are training AI models on non-proprietary broad accessible data, and you see the power of that. Imagine though, training AI models on proprietary data sets not available to others. And I think of three data sets within our company that are very relevant to the discussion. The first is the connectivity dataset, the dataset of Ookla, Ekahau, and that family of businesses. As you know, that business is a data business. The ability to use AI to generate even greater, more frequent, and accessible insights, I think is a big idea. If you think about our RetailMeNot and shopping business, we have a multi-merchant view into shopping behaviors, shopping traffic patterns. So, from a shopping insights and analysis point of view, training AI models on that proprietary dataset is compelling. And then our mod business, as a third example, sits on a universe of XEO data that is very valuable, and in fact, has already been licensed. So, basically the largest AI model we've all heard of, has licensed mod data already. So, we're actually in the business and we think that has long-term and interesting potential. So, both generative AI and proprietary data that AI models can be trained on.