Repeating a costly mistake?

Repeating a costly mistake?
The good old times? People reading newspapers on a New York subway by WYNC under Creative CommonsAttribution 2.0 Generic

Welcome to the sixth issue of my monthly newsletter

I'll be sharing analysis and short stories about digital transformation, practical recommendations, or recommended reading on this platform.

This time let's explore an ongoing rift in the media industry as publishing executives debate how to handle AI and it's media-generating capabilities. They seem faced with a tough decision between making a deal with the devil or fighting tooth and nail against them. As the media industry is faced yet again with technological disruption, the question arises: are executives about to repeat a costly mistake?

Please enjoy!

For the last decades media companies around the world, more specifically print journalism, have struggled with the rise of the internet. The technological revolution brought about by easy access to information and new ways of creating and distributing content from search engines to social media have thrown an industry into crisis and turned the career path of journalist, steeped in prestige and purpose, into a high-risk, low-reward choice. Looking back, one could argue that the strategic decision to provide free access to journalistic content by publishing it on the internet to secure the role of information provider was a gigantic mistake.

Around 30 years after the internet shock, the media industry is in the middle of a second shock brought upon by AI, specifically generative AI like large-language models and other models able to generate media content, from text to images, from sound to videos. Where the internet primarily asked the question of distribution of journalistic content and opened the field to who could generate content, generative AI is fundamentally asking the question of content generation. Media companies have been experimenting with AI for a while, mostly on very structured and shorter news items that would lend themselves easily to being automated like handling of sport results or financial reports by companies. But it is easy to see, how you could think about using generative AI to assist or even wholly create longer forms of content.

Humbled from ignoring the internet for a long-time, media companies are now actively engaging with tech-companies. Rightly afraid that after distribution channels they might also lose their position as content creators, the big names in media are trying to figure out how to handle the new giants on the block, in particular Open AI. At first sight, there are promising signs. Media companies have started coming up with their own charters and codes of conduct on how to responsibly use artificial intelligence in their business and the media companies have something the AI companies desperately need to improve their models: high quality data. Human created content by itself isn't necessarily worth a lot, as Google had to learn the hard way after apparently relying a bit too much on content from the online platform Reddit. But content from prestigious media companies, created by capable journalists and adhering to certain standards for quality would be great training material for AI companies.

No surprise then, that several big names have struck deals with Axel Springer taking the lead and other giants like News Corp and the Financial Times following suit. While the details of the deals differ, they all follow the basic idea of cooperating with AI companies. There are others like the New York Times that have taken a more confrontative approach by suing OpenAI and Microsoft over copyright issues.

This raises the question: what way forward for media companies makes sense in the age of AI? Have media executives learned from their past mistakes (or the mistakes of their predecessors) and is striking a deal the right way to go? There are several arguments to suggest that they are indeed about to make another costly fumble.

The first argument is a rather cynical one. It assumes that the game is lost anyways so it is only about making at least some money while you still can. In this line of argument, tech companies like OpenAI will find ways to get access to data no matter what media companies do. So those deals are a way of getting at least something in return. But clearly, this "hostage situation" does not sound like a sustainable way forward for the media industry.

The second argument is the idealistic reaction to the first one. Media companies and journalists should fight against AI! Arguing mostly from the perspective of journalists, Hamilton Nolan makes the point that cooperating with AI companies would be equivalent to "feeling pleased with yourself that you made five bucks selling your house keys to some burglars". But what's the alternative? Legal instruments such as the NYT court case, protection of journalists and new regulations. Can that really work especially given that those instruments take time, a resource the media industry doesn't necessarily have?

Whatever your position on these demands might be, there is another argument to be made that at least casts some doubts on the value of those deals. And that is that in the current market environment for media companies, business decisions seem mostly driven by cost. From that perspective, generative AI sounds like an interesting option. Why hire expensive journalists that also need days off when you could have an AI newsroom producing content 24/7 and highly personalized to your target audience? Leaving aside the complicated question of whether this would be desirable, I'd argue that this perspective is flawed because you are only looking at one side of the equation.

Clearly, AI has an impact on the costs, but what about the profits? While the business models of AI companies are not yet clear, the same is true for media companies that are struggling with their heavy reliance on advertisement. I don't see how media companies could claw back advertising market share in a market where gen AI turns every individual into a marketing agency. But what could the potential for additional profits from AI be? Well, in a future of not only abundant access to information but a truly staggering abundance of content, generated by AI, the today unprofitable business of journalistic practice might suddenly become valuable. The problem already today is not that we do not have information but whether that information is reliable. Generative AI is only going to make things worse in that regard, increasing the value of trusted sources and content.

So where does this leave us regarding the question of how to deal with AI? Any decision on this should take into account the following points:
First, AI as a technology goes beyond the current hype of generative AI. Try to understand the inputs and outputs of that technology to determine it's value for your business and what leverage you might have.

Second, be wary of rule-breaking masquerading as innovation. The legal frameworks in place still apply and there is no reason for you not to seek enforcement of those rules, e.g. against scraping.

Third, ensure that any use of technology isn't purely driven by either it's cost-saving or profit-generation potential alone, always look at both sides of the equation.

In sum and in the bigger picture, I see the current struggle in media on how to cope with AI as an opportunity to rethink the value and core business model of the whole industry. Let's hope, executives use the opportunity wisely. Otherwise the rift between journalists and publishers will grow further with negative consequences for media as a whole as the recent developments at The Atlantic showed: On the day of the publication of an article decrying collaboration with AI companies as a Devil's Bargain, the mother company announced a deal with OpenAI...

Subscribe to this newsletter and follow me on social media for more publications and thoughts: