Just this week, Reddit signed a long-term partnership agreement worth $60 million per year with Google, allowing the latter to access Reddit user content to train AI models. As one of the largest sources of real human experience data on the internet, Reddit has a unique and vast real-time human conversation archive, providing valuable resources for search, AI training, and research. 🤦♂️
In response, Automattic, the parent company of Tumblr and WordPress.com, is in talks with OpenAI and Midjourney to sell user posts for training AI models. Automattic plans to release a feature setting that allows users to opt out of data sharing with third parties, and they have already collected all public posts published on Tumblr from 2014 to 2023.
Similar to content platforms like Reddit and Tumblr, there is now a new revenue model of selling user data for training AI models, in addition to advertising. What's more, it is a subscription model that can generate continuous revenue. 🦥
So what should we do as individuals? Are we just becoming human mines? 🤔 I searched on Google and perplexity for a long time and found an interesting new project called Wynd Network.
It is a startup company focused on the field of decentralized artificial intelligence, and its Chief Technology Officer said, "We want to make public web data more accessible to open-source AI projects, and decentralization is the only way to achieve this goal in a way that is ethical and efficient." Wynd Network launched a decentralized network resource sharing platform called Grass in June, which is currently in the testing phase. 🤡
Let's focus on this Grass. Its mechanism is that users sell their unused network resources through Wynd Network to companies, laboratories, and other institutions. The buyer companies of Grass seek unused network resources to access more diverse IP addresses for market research or tasks such as web scraping and AI training. Users provide their idle bandwidth resources through Grass and receive points granted by Grass. Users will be able to earn up to $30 per month by selling bandwidth.
Anyway, this can be considered as support for the world data of open-source AI models! 🙋♂️
Let's try it out. Grass currently only offers a Chrome extension account for desktop computers, and the mobile version will be released later. Go to Getgrass.io, register an account with your email, and then add the Grass Extension browser plugin.
If you encounter the need for an invitation code to register, you can try filling in "WMn8Hz3pJSEm8in".
Log in with the email account registered before in the plugin interface. When it shows "Connected," it means the connection is successful. Keep your Wi-Fi connection on to continue earning Grass points. Please note that you need to turn off VPN to earn points. The user's point mechanism is not fixed and is influenced by the geographical location and network environment of the IP.
After reading a lot of materials about AI data collection, Grass is one of the few projects that I think has good prospects, but I'm also afraid it will develop into a P2P torrent player. The issue of data security needs further confirmation. So, I only recommend everyone to try it out as a reference for product design or technical architecture design.
Currently, Grass is only available on the web, and support for Android will be added in the future, which needs to be used with a browser extension.
Welcome to try it out.
That's all, thank you for reading.