Generative Models are built on consuming as much data as possible. Unfortunately, it is not always clear to end-users how that data is being used, or sometimes whether or not it is being collected. Many of us have become accustomed to meeting over Zoom. But when was the last time you stopped to read Zoom’s terms of service? Earlier this year Zoom changed its terms to include that by using Zoom, customers “consent to Zoom’s access, use, collection, creation, modification, distribution, processing, sharing, maintenance, and storage of Service Generated Data” for purposes including “machine learning or artificial intelligence (including for training and tuning of algorithms and models).” In other words, by clicking “Ok” we were agreeing that Zoom could feed our vast amounts of data into its generative AI models. It is not exactly clear how this data would be used, and what would be done to protect personal information.
After users noticed this change in terms and responded with frustration and outrage, Smita Hasham, Zoom’s Chief Products Officer, wrote a post on Zoom’s blog (https://blog.zoom.us/zooms-term-service-ai/) stating, “We will not use audio, video, or chat customer content to train our artificial intelligence models without your consent.” That post was edited, and now reads “To reiterate: Zoom does not use any of your audio, video, chat, screen-sharing, attachments, or other communications like customer content (such as poll results, whiteboard, and reactions) to train Zoom’s or third-party artificial intelligence models.”
Unfortunately this is just one incident of how your data may be used to feed generative AI model’s need for “big data.” When you are asking the friendly chatbot on a company’s website for help in returning a pair of shoes, it is likely collecting data that will be used to feed the machine. And when you are chatting with the helpful customer representative online, remember that it might be a chatbot gathering your information.