What does OpenAI intend to do
Why is OpenAI partnering with these media companies?
Obviously, by doing so, it can obtain authorized training data to build powerful new AI models whose writing abilities will be no less than those of Wall Street Journal reporters.
OpenAI's purpose is to improve ChatGPT's performance and ultimately hopes to commercialize these tools and sell them to these media or other media in the field.
For digital media agencies like Vox that produce video content for YouTube and licensed documentaries and series for Netflix, OpenAI can use their works to train generative AI video models like Sora.
Why is OpenAI willing to pay?
Previously, OpenAI had unabashedly scraped almost all public posts on the internet to train large models.
This move was strongly opposed by artists, creatives, and even media companies like The New York Times.
This caused OpenAI to no longer insist on the stance of "legally scraping public data to achieve transformation", and they were challenged on data ethics.
Therefore, OpenAI launched a code last year that website owners can add their own sites to a list to prevent it from scraping websites for training.
OpenAI also recently announced that it will create a new product - Media Manager, which artists, creators and publishers can use to mark works they intend to or have already published online, indicating that they do not want to see these works crawled by AI tools and used for training models.
However, this won't be realized until 2025, and content creators may not necessarily buy into it.
Therefore, for OpenAI, which is in the limelight, paying publishers to keep quiet and accept AI scraping and training is not a losing business.
On one hand, it can get itself out of trouble and obtain the needed data, and on the other hand, it can also give an account to investors and users, showing its compliance with copyright laws and ethical norms.
However, content owners have not received real returns in this process.
Can media publishers really benefit
Publishers have all announced OpenAI content licensing agreements without exception, and they did get something, the most important of which is not money, but "position".
Specifically, almost all publishers pointed out that ChatGPT will display their articles in its output.
Therefore, if users input "summarize the latest tech news", it may display article summaries from Business Insider, The Verge (under Vox), The Wall Street Journal or any other publication included in the deal, along with source links.
This is just a possible scenario, and the exact agreements or technical documents have not been publicly shared.
Moreover, it is still unclear how ChatGPT will use content from the media. If it adopts a "robotic" style to summarize based on the original text, it may erase the artistry created by the original author.
Also, since users have already read summarized news on ChatGPT, they won't choose to visit the website that first published the article, thus causing these publications to lose traffic. For publications, this means losing paid users or commercial value.
This is why industry veterans like Jessica Lessin, founder of The Information, former Gawker reporter Hamilton Nolan, and former Vice reporter Edward Onswego, Jr. all pointed out that publishers are bearing cruel consequences in deals with OpenAI.
After all, if readers are pursuing pure information, and ChatGPT provides them with pure information, what's the use of visiting the original media, let alone paying for subscriptions.
Users will choose to become ChatGPT Plus, paying OpenAI $20 a month, instead of patronizing media that produce content.
History repeats itself
What's happening now reminds people of when Google News was just launched. That was in 2006, when the number of users and popularity of social platforms like Facebook and Twitter began to grow and quickly became the main source of referral traffic for publishers.
This situation has basically remained for the past 15-20 years. However, due to the management and constant algorithm adjustments by the tech giants behind these platforms, traffic has fluctuated.
When a tech platform suddenly changes its algorithm, causing audiences to disappear, websites that have invested too much energy in a particular platform or strategy quickly find themselves at a loss.
Of course, changes are still happening, and it can be said that the biggest change is now in front of technology platforms and publishers: generative AI.
As Google puts its erroneous AI summary results at the top of search result pages and pushes down direct links to publishers and news articles, more and more people may choose to adopt ChatGPT as a news source or aggregator.
Executives of news publishers and parent companies may feel cornered: the game is changing again, AI is coming and replacing traditional ways people get news online. So why not cooperate with the disruptors and try to ride the wave?
However, as the brief history lesson mentioned above shows, tech companies always change strategies and tools randomly and unpredictably, which frustrates media companies.
Although OpenAI is now getting along well with publishers