Is AI The Friendly Skies for Publishers? A View From 34,000 Feet, (Literally).

by Frank Bilotto

If it’s not too soon, I wanted to write another piece on my newly found favorite subject: The impact of AI on Publishers. To clarify the headline,  I just boarded a plane bound for Phoenix. So, for the next four hours, I’ll try to package what I’ve learned so far, the pertinent issues, and what you, the publishers, should consider moving forward. 

If you’ve read my previous articles, you know that I’ve warned publishers to be cautious about AI, and for good reason. AI stands as both a marvel and a disruptor for content creation. While I’ve been tough on the negative impact of AI, in this piece we will consider how you can expand your reach and profit from AI. To do so, you must learn how AI is using your content and the business models associated with it. I assure you that the publishers that make the effort will flourish for the foreseeable future.

The Yonder Isn’t Completely Blue, But it is Wild.

Before I get to the benefits of working with AI technologies, let’s quickly review  the significant risks that publishers may face when collaborating with AI developers. 

First and foremost, don’t ever forget that your most valuable asset is your intellectual property.  Allowing generative AI technologies access to your content opens a Pandora’s Box of potential misuse or misappropriation of your copyrighted material, including unauthorized reproduction, distribution, or modification. The most unnecessary expense you could ever incur is to prosecute or defend copyright infringement claims.

Second, generative AI technologies may not always produce output that meets the standards of quality, accuracy and integrity that you expect, potentially diminishing the credibility and reputation of your brand. When granting unrestricted use of your content to an AI developer, you are essentially forfeiting control over the creative process and editorial decisions inherent in your own content.

Third, permitting generative AI technologies to access your content relinquishes your control over how your intellectual property is used, manipulated and repurposed. You should also be aware of the  potential for biased algorithms that could affect the use and consumption of your content.

Finally, as AI-generated content competes with and devalues the market value of original content, (and it most certainly will), unless appropriate compensation mechanisms are implemented, you will see declines in your traditional revenue streams from your intellectual property.

Feel Free to Move About The Cabin

Are you scared, yet? We’re passing through a patch of some of the roughest turbulence I’ve ever experienced, and I’m still convinced that the survival risks at this very moment are greater to you than they are to me. But before you panic and reach for the exit door, I want you to consider the benefits of working with AI technologies. By prioritizing intellectual property protection, quality control and revenue sustainability, you can make informed decisions about whether and how to engage with generative AI technologies in a manner that aligns with your company’s objectives and values.

It’s Time to Earn Your AI Wings

The fundamental reason you should be licensing your content to traditional aggregators is to expand the reach and consumption of your content, thereby increasing your gross revenues. The reason you should be licensing your content to AI developers is exactly the same. As with any new venture, you have to think strategically. 

At CLI, we’re looking for ways to limit the risks associated with AI, while taking advantage of what we see as low hanging fruit. For example, we are encouraging our publishers to engage with aggregators that are developing AI technologies to improve the discoverability of their content.

Today, machine learning algorithms can tailor content recommendations based on user preferences, interests and browsing history that enhances user engagement and satisfaction. Major aggregators are developing AI that will automate various aspects of content creation, curation and distribution, leading to increased efficiency of discovery by users and greater exposure for publishers. We see this use of AI as a potential windfall for publishers with high quality content that is often lost in a pool of content that may include publishers with bigger brand recognition. 

So, our publishers are executing agreements that give access to machine learning technology, but limit the use of AI to enhance search functionality. 

Today, we believe that unless royalty or license fee arrangements for use by generative AI technologies are substantial, the risk to you is greater than the potential reward. But we’ve begun to explore and develop the most lucrative and risk averse ways for publishers to be paid, so that we’re prepared when the winds shift. 

Coffee, Tea or Milk (Might be Outdated, but it’s the Best I could do)?

The flight attendant just offered a wide variety of soft drinks and alcoholic beverages to me, which reminds me of the ever-growing variety of royalty arrangements and fee structures for giving AI technologies access to your content. These payment arrangements can vary depending on several factors, including the type of content, the scope of usage, the commercial potential and the negotiation dynamics between publishers and AI developers.

In previous articles, I’ve taken the position that the most predictable revenue stream is for publishers to negotiate upfront licensing fees for the right to use content in generative AI applications. Licensing fees can be one-time payments, recurring fees based on usage, or a combination of both.  The duration of the license, the perceived value of the content and the commercial opportunity for the AI application should factor into the negotiations. The risk to both parties in license fee agreements is the accuracy of the valuation of the content.

An alternative that mitigates license fee risk is for publishers and AI developers to share in the revenue generated from the use of generative AI applications, based on a percentage of gross revenue, net revenue, or other predefined metrics. Revenue sharing agreements incentivize both parties to maximize the commercial success of the AI application, while ensuring that publishers receive fair compensation for the use of their content. As a publisher, your negotiation tactic should be less focused on the percentage of the total content pool that is made up of your content, and more focused on the value your content brings to the performance of the generative AI. 

I’m beginning to lean into the idea of publishers offering their content to AI developers through subscription-based models, where developers pay a recurring fee for access to a catalog of content. Subscription models provide developers with flexibility and scalability in accessing content for generative AI applications, while providing publishers with a predictable revenue stream over time. My only issue with this is breach or termination of the agreement, because once the content is part of the AI technology, it cannot be removed.  So, subscription models should begin with an up front fee for the archived content that is large enough that you won’t be concerned if the agreement terminates. Then you’ll negotiate a  subscription fee for newly added content that should consider the value of the shelf life of the previously added content that is aging in the catalog. 

We’re not there yet, but at some point, AI technologies will be able to accurately track the volume and usage of specific content in generative AI applications. When the technology gets there, I am certain that I’ll be a proponent of usage-based fees that are structured on a per-unit basis. I’ve promoted content by the slice models since the early 2000’s. Once the technology is reliable in AI, you will be able to generate revenue based on the actual consumption and impact of your content in AI applications. I’m convinced that this will become the dominant model in the future, as it is the most equitable to both parties.

Frequent Flier Miles are Ubiquitous. Words are Not.

At CLI we’ve been approached by a number of AI developers offering to pay our publishers by the word in exchange for unrestricted use of content. This is a long standing practice for the way publishers have paid their writers. Clearly,  the most important component of this arrangement is to ensure that you are fairly compensated for the use of your content by generative AI technologies. You risk undervaluing your intellectual property rights and the contributions of your content creators.

Payment by the word for content generated by AI technologies incentivizes quantity, which may not align with your goals as a publisher. Additionally, this model may not provide you with sustainable revenue streams in the long term, especially if the market becomes saturated with low-quality, commoditized content generated by AI technologies. In case you haven’t noticed, it’s already happening.

While payment by the word may offer a straightforward metric for compensating you for content generated by AI technologies, it may not fully capture the value, quality, and originality of your human-authored content. The position of CLI is that unless the price per word you are offered dynamically changes your financial statements, walk away until further notice.

There are at least a dozen other ways to structure licensing fees or royalties, including various performance-based incentives or hybrid models. You’ll implement the one that provides you with the most comfort. That comfort typically will come from factors other than money.

Trust, but Verify.

I’ve always said business, like life, is about relationships. Publishers should thoroughly evaluate the credibility, reputation and financial stability of the aggregator or AI company. Work only with established entities that have a proven track record of success.

This may seem obvious, but it needs to be stated. You should negotiate clear and detailed agreements with the aggregator or AI company, outlining the terms, conditions, and timelines for usage, dissemination and future royalties. 

You should demand that AI developers be transparent about the use of machine learning technologies and provide clear explanations of how algorithms are trained, evaluated and deployed. Additionally, you must address issues of bias and fairness in machine learning algorithms to ensure equitable access and use of your content

Often overlooked in licensing agreements are risk mitigation strategies and contingency plans to address potential challenges or uncertainties related to usage and royalties. When licensing to AI technologies,  these provisions become even more important to include.

And, don’t forget to clarify ownership rights and intellectual property provisions in the agreement to protect your content and creative assets. It’s essential to define the scope of usage, licensing terms, and attribution requirements to ensure that you retain control over your content and receive proper recognition and compensation for its use.

Finally, you should consider the potential for long-term collaboration and partnership with the AI company beyond the initial agreement for future royalties. Building a mutually beneficial relationship based on trust, transparency, and shared goals can foster innovation, collaboration, and sustained success over time.

My plane just touched down. So, even if I could, I won’t be writing a pithy conclusion, primarily because even though my literal plane has landed, your metaphorical plane is ready for take off. Besides, this article is already too long. For those of you still reading: Safe Travels!