We appreciate your continued use of this site.

General
We appreciate your continued use of this site.

We reported in May back to a deal that Reddit was doing in an unnamed AI costume at the time to open access to user posts for AI training of about660million, although that deal turned out to be with Google, currently Reddit is inking another arrangement with OpenAI.

In a blog post, Reddit outlined the deal, but did not reveal how much money was involved. However, there seems to be a difference between Google and OpenAI deals. At Google, there was an explicit reference to AI training.

However, the OpenAI agreement states that "we are talking about allowing access to real-time, structured and unique content from Reddit." This will allow OpenAI's AI tools to better understand and showcase Reddit content, especially on recent topics."This means that OpenAI's models can refer, link, or quote Reddit posts, but in return for revealing to Reddit posts the training of future versions of OpenAI's ChatGPT model

, Reddit"can bring new AI-powered features to redditors and mods."

For what it's worth, OpenAI posted the exact same statement on its website. Now, save for one additional comment at the bottom with a small font. 

"OpenAI Disclosure: Sam Altman is a Reddit shareholder. The partnership was led by OpenAI's COO and approved by an independent board of Directors.

Well, of course Altman is a Reddit shareholder. It is incompatible with the broad sense of corporate and technological supremacy if the key figures in AI are not even involved in major social media.

Anyway, it's not hard to see how such a deal makes sense. The training value is obvious enough. There is not so much historical content published on the Internet. It has already been shaved off by existing models, and various AI outfits are said to lack new content to train models.

Indeed, OpenAI, desperate for new input data, reportedly transcribed millions of hours of YouTube audio into text to help train ChatGPT4.

At the same time, its usefulness is limited by the inability of many LLMs to access up-to-date and live information. In many models it is as if the world stopped at any cut-off date of a year or two.

In addition, whatever you think about giving Reddit posts specifically access for AI models, for training or reference purposes, such a deal is at least possible if you post on Reddit, you can't wonder what the post will be used for.

Similarly, these Reddit deals with 2 of AI's biggest players show that the latter is willing to pay for access to content rather than using everything without permission.

If you want to know if you can advance on all the data they access, or if you want a broader understanding of how AI affects online content when owners are aggressive and likely to sue, check out Jacob's new extensive overview.

Categories