Kevin Bartus’ Post

I guess this will prove if Artificial Intelligence is an oxymoron.

A "Preferred Publisher's Program"- possible response to the spate of lawsuits (NYT being the most prominent) by publishers. Wondering how this will be different from our current search engine model? Instead of relying on relevance of responses, we will be back to consuming biased/preferred data, determined by the amount content providers pay (akin to ad spends with Google). Especially with the upcoming announcing of the chatGPT powered search engine. Round and round.

TIME magazine and OpenAI have just announced a multiyear content licensing deal. This partnership will tap into Time's rich 101-year-old archive and real-time content to train OpenAI's AI, enhancing its ability to answer user questions. OpenAI, in turn, will credit Time in its responses and link back to the original articles on Time.com. Not only that, Time will leverage OpenAI's technology to develop new products and enhance AI-driven journalism. This collaboration mirrors OpenAI's recent content deals with The Atlantic magazine and Vox Media, pointing towards a trend in the industry. #Time #OpenAI #AI #ContentLicensing #Innovation

Interesting new content deal from OpenAI: as reported by Axios, Time Magazine has provided its content library to OpenAI. Content for training isn't all that interesting. What *is* interesting is the reference in the article to "access to Time's real-time content to help answer user queries as news unfolds," in other words, mostly likely RAG through an API. I think this is the future of media syndication and portends a symbiotic co-existence between LLMs and content. Solves AI's "what's new?" problem and solves media's syndication problem.

The Financial Times and OpenAI have partnered up for a content licensing deal. - FT is sharing its content to help OpenAI develop advanced AI tech for text, images, and code. - ChatGPT will now provide FT article summaries to its 100M users. - This is OpenAI's fifth major partnership following deals with the Associated Press, Axel Springer, Le Monde, and Prisa Media. - Deals like this are more common following the NYT suit accusing OpenAI of using millions of articles to build the models underlying ChatGPT. https://lnkd.in/etGGSiS9

#ChatGPT will now train on #FinancialTimes contents. This highlights how #GenAI highly depends on input data. We will see more such announcements. Whilst positive for corporate users of #OpenAI what does it bring to the reader other than a re-hash?

The Financial Times has licensed its content to Open AI. This is quite a big moment. It joins publishers #AxelSpringer, #AssociatedPress and #LeMonde amongst quality producers who have taken money from the Microsoft co-owned Chat GPT producer. #NewsCorp is also in talks to do something similar. It puts the #NewYorkTimes into a small category of willing resisters - the other major global title that has, like the FT, monetised its output on a subscription model outside the clickbait relationship with particularly social channels that has been challenging for many newspaper groups. I bet the NY Times conclude their current principled court action - with a settlement and a training data deal, like the FT. But it does raise other interesting questions too around the relationship between #journalism and #LLMs * If articles are accessible via ChatGPT how will they be monetised and how will the FT not cannibalise its powerful subscriber audience ? * What is the pricing model for journalism as training data and does it contribute meaningfully to the costs of production * Within the walled garden of #LLM access to #FT material how will potentially brand-damaging hallucination in outputs be prevented ? * Is it a win for journalists for their output to be training data ? Or is it vindication of the value of high quality output that it's needed in a global environment of increasingly scarce new high quality material for training ? There is only so much value that LLMs can extract from vectorising the content of #Reddit threads and #youTube comments. * Finally, if responsible media outlets did NOT allow themselves to be trained on, would right-wing world views start to take over the LLM outputs as sites like Breitbart stay open to crawlers ? Sounds exotic but it’s a real risk.

ChatGPT enhancements marks a significant step towards transforming how we access and use information, potentially reshaping search dynamics and providing new opportunities. Intero Digital https://lnkd.in/gwDYGHZY

Time is on the AI side. Time Inc. has struck a multiyear content licensing deal and strategic partnership with OpenAI, Axios reported. The collaboration aims to expand access to Time's content for younger and more diverse audiences worldwide. Under the agreement, OpenAI gains access to Time's archives spanning 101 years to enhance its large language models, notably ChatGPT. The partnership also allows OpenAI to utilize Time's real-time content to provide timely responses to user queries. Additionally, OpenAI will attribute Time in its responses and link back to the original content on Time.com. In return, Time will leverage OpenAI's technology and tools to innovate new products for its audience. The collaboration aims to enhance journalism delivery and develop AI-driven news experiences. Financial terms were not disclosed, but previous OpenAI deals have included compensation for publishers. Mark Howard, Time's chief operating officer, highlighted that the partnership aligns with their mission to provide trusted information globally. Brad Lightcap, OpenAI's chief operating officer, emphasized the importance of proper attribution to original sources in supporting reputable journalism. #time #ai #chatgpt (Source: Axios)

OpenAI strikes content licensing deal with Financial Times for an undisclosed amount. This comes after similar deals with NYT and Axel Springer, individually worth multiple tens of millions $$ p.a. ChatGPT users will soon receive summaries and quotes from FT/NYT/Springer journalists, as well as links to original articles. It's an interesting development and looking at potential paths forward, it seems that an 80/20 Pareto scenario is the most likely outcome: Few partnerships with major publishers to include majority of relevant content from the most credible sources. One step closer towards copyright and IP protection in the age of AI.