Mark Stenberg’s Post

AI's capability depends on the content it can access. Now OpenAI plans to pay publishers to access the latest content while giving them better exposure in its responses. Now if a publisher signs up, they - Get paid - Improve their brand exposure - Shape future responses to their Liking AI Bias is already a concern expressed due to biased training data but I expect it to move a further down as many publishers won't sign up(https://lnkd.in/gNU7AwM3) or not eligible. Those who sign up will have an edge in Search and brand perception over others. It will be interesting to see how OpenAI manages the program, while trying to reduce bias in ChatGPT's responses https://lnkd.in/gWSps5_5 #AI #Bias #Technology

One of the big questions around Large Language Models today is whether OpenAI, Microsoft and Google should pay creators for using their content to train AI models. Although publicly, most of these companies say their usage of content from the internet to train models is 'fair use', a few of them have struck deals with publishers. This article on ADWEEK shows some insights into how OpenAI pitches publishers on these partnerships. #ai #genai https://lnkd.in/g6nGnKjJ

According to this AdWeek report, OpenAI is offering publishers both payment and prominance in chat results: Not pay-to-play but get-paid-to-play (and btw, don't sue or lobby). There are issues: Credibility and bias in chat results and, most important, the exclusion of small, Black, Latino, community, startup media, which leads to more bias in results. Rather than all these early-birds-get-the-worm deals with AI companies, the news industry -- and the culture -- would be better off creating a collective API for news to license to AI companies. But news companies never collaborate, especially not with each other. The big guys get bigger; that's what's happening here. https://lnkd.in/e5wb2eSr

A leaked pitch deck from OpenAI indicates that the generative AI company may be willing to pay to access high-quality data from publishers to train its models on. In some ways, it vindicates all the people who said that AI can't replace human-generated content when it's authentic and thoughtful, but I get the sense that there may be some trouble still with this kind of model of licensing. Granted, for a generative AI tool to provide up-to-date information, it needs up-to-date data, but will the information still be as useful when publishers can pay to have their content listed as a priority in ChatGPT responses? And what about writers; do they get to see some of the payments their work gets? AI tools are going to become a lot more prevalent in the way we work, but it's critical that we, as a society, try to implement it in a way that minimizes harm to those who will be displaced by it, and recognize the value of people who make the generative AI tools as advanced as they are, whether it's a data scientist, an engineer, or a writer who provides the information needed to create such models in the first place. #ai #generativrai #chatgpt #openai #publishers #writers #data #management

OpenAI's content deal controversy comes down to the balance between AI's potential value and threat. This serves as a mirror for #marketers navigating the #GenAI era and offers insights for platforms like #Google facing AI backlash. Is AI a looming threat to content and creativity, or the next stage of evolution? I believe that AI's immense value can overshadow fears through genuine transparency and collaborative efforts. Key Takeaways: **Value**: Monetization, Data, and Traffic - Publishers stand to benefit from substantial revenue and attribution value, contingent on clear pricing structures. - Marketers can leverage AI for enhanced efficiency, data insights, and branding opportunities. However, learning from OpenAI's model, platforms must prioritize collaboration over competition to maximize value. They need to avoid rolling out AI without methodology, attribution or even accuracy (Looking at you Google and others). **Threat**: Exploitation & Replacement - The fear of exploitation looms large as AI leverages content without consent, placing publishers and marketers in challenging negotiation positions. - The ongoing concern revolves around AI potentially reaching a level where it could replace the human touch entirely. In the face of transformative technology, immense power demands profound responsibility. While regulations catch up, AI companies and platforms must prioritize customer inclusion, actively pursuing collaboration and partnerships for a sust ainable future. #content #strategy #digitalmarketing #SEO #openAI Disclaimer: This post was not written by, but optimized with AI 🤣

News has for generations been most peoples window to what is going on locally and across the planet. The last decade it has fallen under a lot of scrutiny for being biased or focused on entertainment more than news. But at the end of the day it was people reporting on events. This might be changing. Both Elon Musk with Grok/X and OpenAI are now working deeply to start creating the news content using AI based on published information and from social media. Where can this lead to? Will these type of tools start to eliminate the humans in the creation of news content? How does this impact investigative reporting? And bias, will this now be influenced by the model creator? Many questions arise from this but change is coming. #ai #news https://lnkd.in/gimzr3gG

Great to see news publishers get some value out of their dusty archives - Financial Times licensing to OpenAI in the latest instance! Deals like this make a lot of sense for content that gets stale quickly (like news and other current affairs) - especially if they have 1) fixed terms for archive licenses, 2) subscription components for new content, 3) attribution through RAG, and 4) referral to original content. However, they make much less sense for long-lived content - such as education, academic, technical, legal, and other reference collections. Reference content stays relevant for much longer, and so its value decays more slowly - changing the implications and economics of licensing such content to LLMs. Because there's less need for up-to-date content, more of the value is in the archives, creating far higher engagement and revenue risk for these publishers from licensing to LLMs. It can still be done - safely - but requires a different approach to deal strategy. At Outsell, Inc. we've been working with a wide range of publishers on AI licensing strategy, with clear best practices emerging. We've published a basic licensing strategy framework here: https://lnkd.in/ehnJYRqv.

Question: can AI Agents qualify in some ways as "Audiences"? I realize the obvious knee-jerk answer is "No, we aren't transacting on them." - yet. This may be worth further consideration for many companies in our industry. Given that AI agents are making rational decisions based on their perceptions and data to produce optimal performance and, ultimately, meaningful results there are sectors of the media and ad industry that should reconsider the impacts and opportunities for their businesses. Everyday, we're hearing more about Generative AI reshaping industries, particularly the media landscape - for some, an existential threat. AI Agents and Retrieval Augmented Generation (RAG) are constantly scraping a publisher or streamer's information, data and content, processing and optimizing the perceived value, and rendering it or part of it in some way. As users scale these AI products they will continue to directly impact most businesses across the web. So, with all of this in mind, I'm surprised we're not seeing more position papers, consortiums or proposed Gen AI products or "potential standards" coming out of the Verification and Measurement sector of our industry. I'd think that they have even more to lose than publishers and content creators. I know this sounds really abstract right now but shouldn't the MRC at least consider, discuss and have an evolving position on what may eventually constitute as an "Accredited environment" within some forms of Generative AI? Thoughts?

As it says in the quote from John Ridding below, this is about far more than the #FT. “Apart from the benefits to the FT, there are broader implications for the industry. It’s right, of course, that AI platforms pay publishers for the use of their material. OpenAI understands the importance of transparency, attribution, and compensation – all essential for us,” said FT chief executive John Ridding. It would be fairer to say that #OpenAI is learning the benefits of transparency, attribution, and compensation. This has been a hard-won lesson for OpenAI. Its earlier approaches paid scant regard to copyright, attribution and compensation. The lesson OpenAI seems to be learning is that without quality training inputs, the output quality also takes a hit. If the output quality is poor then the overall value is reduced. I'd love to know the terms of the FT deal, because I suspect the economics will be complex. The FT has, for years, done a good job on pricing and value (which is why I subscribe). How this will play out in the world of genAI is going to be very interesting to see. This is AI #economics in the making. It underlines the symbiotic relationship between all ages of quality media. I just looked up when the FT was founded. The answer came from #Wikipedia (in a side window on #Google): it launched in January 1888, as the London Financial Guide, before rebranding just a month later. See, nothing new in experimenting! Asif Javed, Joanna Levesque, Tim Harford, Lucy Watson

Brand Finance's new analysis has found that global intangible value has increased by 8% from $57.3 trillion in 2022 to $61.9 trillion in 2023. The Financial Times' collaboration with OpenAI to train AI models using its archived content reflects this shift and increasing recognition of intangible assets' value in the digital age and how companies can monetize them. I am excited to see the Financial Times pioneering innovative partnerships in the realm of #AI and journalism. A two-sided puzzle I have. What bandwidth will OpenAI have to do these types of license deals? Second, should data subscription businesses have a licensing deal with OpenAI as an opportunity or a threat?