The question is now personally pressing, as the FT last week reached a deal with OpenAI to train large language models on our content. Like many news organisations, the FT will now receive payment for its intellectual property and also get attribution when the content is surfaced.
On the one hand, you can argue that it is better to be paid than not for something that Big Tech is already doing for free, which is hoovering up journalism online to train LLMs. (FT journalists have not been privy to how much our company is being paid, or any of the financial details of the contract with OpenAI, so I can’t say whether it’s a good deal or not.)
On the other hand, I worry that we are about to see a repeat of what happened in the mid- to late-1990s, when media companies bought into Silicon Valley’s line that “information wants to be free” and didn’t take a hard line on protecting the copyright and value of their content.
As I outlined in my second book, Don’t Be Evil, and tech journalist Steven Levy covered in more detail in his book In the Plex, Google in particular simply ran roughshod over copyright in a manner that paved the way for surveillance capitalism in general, including today’s AI applications. As I wrote in my January column on the new lawsuits by news providers against AI copyright infringement:
One recent study by researchers from Columbia University, the University of Houston and the Brattle Group consulting firm . . . estimated that if Google gave US publishers 50 per cent of the value created by their news content, they’d be shelling out between $10-12bn annually. As it is, the New York Times — one of the largest news publishers — is getting a mere $100mn over three years. Now, AI is poised to make even that asymmetric relationship look good. When you ask a chatbot such as OpenAI’s ChatGPT or Google’s Bard a question, you don’t get sent to a creator’s website. Rather, you are given the answer directly. Users remain in the walled garden of whichever Big Tech company owns the artificial intelligence platform. The fact that the AI has been trained on the very same copyrighted content it aims to bypass adds insult to injury.
The news business has of course been hollowed out over the past two decades, as a handful of digital giants took the bulk of value generated by online content. This was particularly true at the local level; indeed digital monopoly power has led to so much destruction of local news that Big Tech giants like Google and Facebook, now Meta, started funding the creation of more local news content because the lack of it was harming their own business model. Marx is chuckling from the grave about all this, no doubt.
These days, there are two ways to be in news. You can be the 1,000-pound gorilla, like The New York Times, which has reached escape velocity with more than 10mn subscribers. Or, you can be a high-end premium subscription organisation, which is the FT model. I’ve always thought that the latter was the better proposition, and an easier one to defend. It survived the last round of digital “creative destruction” over the past 20 years. How will it fare in the next?
I honestly don’t know the answer to that question. What I will say is this: I think AI is clearly a bubble. As I wrote in March, that doesn’t mean that something useful won’t come out of it eventually (the late-1990s digital bubble yielded tons of froth, but also millions of miles of broadband cable). But for now, the valuations and the wildly simplistic and optimistic storyline reminds me very much of the mid-1990s, when whatever Silicon Valley said just seemed inevitable. If you questioned it, you were a Luddite. You didn’t “get it”. Most journalists and executives were too cowed by tech titans’ confidence game to really think clearly about how they were giving away their own product for free.
This time, I guess we are getting something for the product. But I’m also very sceptical that AI will seriously benefit anyone aside from Big Tech in the short to medium term. Remember when SAP became red hot in the 1990s by selling everyone enterprise software licences? SAP’s own share price rose but the huge productivity boom that was promised never really materialised (most economists now see that 1990s productivity blip as a bit of a mirage). In the end, the smart bet was to buy SAP and sell many of its clients.
I fear that the AI narrative today is all too similar. For starters, I’ve seen almost no evidence of AI truly enhancing productivity at scale. There are lots of small-scale experiments that have been pushed by the Big Tech firms themselves, and few of them have been properly vetted or audited. We just take the word of the technology companies themselves. We still continue to buy into the inevitability of this story, how if we don’t get on this new “disruptive”, “innovative”, “world-changing” bandwagon, we’ll somehow be missing out.
But consider an alternative scenario. What if the water and energy consumption of AI simply make it unsustainable? What if the productivity “miracle” never comes to pass? What if AI is, in the end, yet another way to cut human labour costs — be it in media or any number of other white-collar industries — and replace people with a technology that monetarily benefits mainly a handful of large firms? What if AI is really just the latest scam by Big Tech companies looking for a way to keep the music playing as interest rates rise, antitrust actions bite, decoupling shuts them out of the China market, and it simply gets harder in general to make money?
Peter, what say you to all this?
Recommended reading
There are so many great FT offerings at the moment, that I’m going to do an all in-house recommended reading list this week:
-
The FT’s Henry Mance did a long-form interview with economist Joe Stiglitz, who rightly points out that “Trump is what neoliberalism produced”. I must say, I disagree with Mance that the needs of the global south and the US are in conflict when it comes to things like climate change. America’s industrial policy approach to global warming, and in particular Katherine Tai’s calls for a “postcolonial” trade paradigm in which America could offer tech transfers to developing countries in exchange for, say, rare earth minerals and other inputs, is much more palatable than the European carbon tax approach, which makes the global south pay disproportionately for the sins of richer countries.
-
My colleague Simon Kuper looks at the contrasting policy choices and lifestyles of Americans, who have more money, versus Europeans, who have more time. As someone who has lived in both places, I think Europeans are making the safer, healthier bet.
-
If you missed the FT Weekend Festival in the US, you can still livestream some of the events and speakers. It’s a truly amazing event, and I’m so sorry I couldn’t be there myself this year (but I had a good excuse, which is that I’m celebrating my husband’s birthday in Greece — back on the 10th, and back in the saddle with Swamp Notes the week following!)
Peter Spiegel responds
Rana, one of my favourite analytical frameworks in the tech world is the Gartner Hype Cycle. It’s been around since the mid-1990s, but I only learned about it last year when I hosted our brilliant tech columnist John Thornhill on a podcast about the future of the sector (slightly embarrassing disclosure: my wife works for Gartner, so I probably should have known about this earlier).
The Gartner Hype Cycle argues that all new technologies start with an “innovation trigger”, or a new breakthrough that gets everyone excited, and eventually reaches a “peak of inflated expectations” before falling into a “trough of disillusionment”, where all the life-changing promises made about the new technology fails to deliver. I have a sinking feeling that we’re close to the “peak of inflated expectations” when it comes to generative AI.
There’s no doubt artificial intelligence will change the profession of journalism, and the FT is already doing some fun things with it — including beta testing an “Ask FT” generative AI tool that allows subscribers to pose questions to a bot, which answers using two decades of FT content.
But is generative AI about to lay waste to the business model that the FT and other leading news organisations have adopted in the digital era? I suspect we’re headed to the “trough of disillusionment” on that one. News consumers — particularly those that need real-time information to make business decisions — can’t rely on the historic and at times inaccurate information currently provided by AI chatbots. I still think they’ll be coming to ft.com and the websites of our rivals for that kind of news.
The Gartner Hype Cycle also predicts, however, that the “trough of disillusionment” is followed by a “slope of enlightenment” where we figure out how, exactly, a new technological tool is best used. That’s followed by a “plateau of productivity”, where mainstream adoption becomes a reality.
Like you, Rana, I don’t know the answer to the question of what the next round of “creative destruction” will look like. People far smarter than I am are still grappling with how generative AI will affect the news industry. Until we start climbing out of the “trough of disillusionment”, though, I’m perfectly happy to play with the Ask FT bot and dabble in other forms of experimentation before I start fretting that journalism is about to be subsumed by Big Tech.
Your feedback
We’d love to hear from you. You can email the team on swampnotes@ft.com, contact Rana on rana.foroohar@ft.com and Peter on peter.spiegel@ft.com, and follow them on X at @RanaForoohar and @SpiegelPeter. We may feature an excerpt of your response in the next newsletter
Recommended newsletters for you
Unhedged — Robert Armstrong dissects the most important market trends and discusses how Wall Street’s best minds respond to them. Sign up here
The Lex Newsletter — Lex is the FT’s incisive daily column on investment. Local and global trends from expert writers in four great financial centres. Sign up here
Discover more from Today Headline
Subscribe to get the latest posts to your email.