Reddit was considering whether to block Google and Bing search engines from indexing posts on the site in late 2023 According to the Washington Post, the decision was to prevent unauthorized and unpaid use of posts to train AI
Now, Reddit has announced that it has reached an agreement with Google that, among other things, gives Google access to the Reddit Data API "to improve its products and services," including "ways to train models more efficiently" In Google's words, access to this API will allow the company to get "real-time, structured, unique content from a large, dynamic platform"
The deal, which Bloomberg previously suggested was "worth about $60 million on an annualized basis," goes beyond that As part of the deal, Reddit will have access to Google's Vertex AI service, which should improve internal search results, and will also allow "Reddit content to appear across Google products"
Google says this will "make Google products more useful to users by displaying Reddit information more content-forward, making it easier for them to participate in Reddit communities and conversations" Considering the number of people who append the word "reddit" to their searches to view authentic user-generated insights, this could be a very good thing for the average Google user
But the real prize for Google, no doubt, is the vast trove of learning data In theory, the posts and comments written by millions of real people every day will make the generated AI seem more human
But scale is not everything, and compared to literature and magazines, Reddit is in some ways an imperfect sample for training artificial intelligence Grammar is faster, looser, full of memes and inside jokes, and just plain misinformed and male
In contrast, Apple is reportedly seeking multi-million dollar contracts with publishers to train in more formal and factual magazines and newspapers But there are obvious downsides to this, too, at the expense of the way people communicate on a daily basis, concentrating on a small part of the human experience [Because people are realizing that AI means big money and that training data will not be absorbed free of charge without consequences Last year, Open AI, Meta, and Stability AI were all hit with lawsuits from authors who claimed their books were used for training without permission or compensation
Comments