The Future of Multimodal Search

Google used to have an entire team that would stitch together the results from like news, videos, images, and the web documents all into one cohesive search page. When Neural IR, when you have a neural network that can embed different types of modalities and text in different languages into a common embedding space,. All of that different content comes back pre-ranked for you in response to a query. And I think because of that, we're going to see these kinds of like search capabilities that have been basically limited to Google and Microsoft in the past, becoming pervasive across the industry.