The spread of conversational interfaces will change media habits

The good news is AI Bing hasn’t yet taken control, at least that we know of. I have to admit being fascinated by the spate of advancements in artificial intelligence and the potential for remaking a big chunk of the publishing business. It’s coming at an interesting time for the media business that feels on the cusp of a new stage, only the parameters of what that will look like remain blurry. First up, a message from The Rebooting sponsor Glide.

Glide makes a cloud-based publishing platform used by leading publishers like News UK. Glide CEO Denis Haman explains why most publishers should adopt a lean media approach:

Know what makes you unique, and focus on that capability. The rest of your tech? Get it brought to you on a plate – supported, maintained, managed. Why spend an age building a totally blah back-end system when you can gain months or years of insight and income using a turnkey service? When the penny drops with CEOs and CFOs that, for cents on the dollar, good SaaS surpasses homebuilt systems for speed and capability, it’s a catalyst for progress. They see the total cost of ownership – and of opportunity -- when looking at tech. Why do you think people like AWS and Amazon use things like Excel and Powerpoint internally? They focus on making stuff their customers use, not remaking someone else’s stuff where it has no value. That’s the lean publisher mentality.

Learn how Glide works with publishers to build lean media.

The return of conversational media

What has long made the internet unique as a publishing medium is how it toggles between communication and publishing mode. In traditional media, publishing was a one-way affair, but the internet was where the audience shifted to become participants. Moving from “read only” to “read-write” mode gave digital media properties and advantages traditional media didn’t have.

The promise of digital media in the Web 2.0 days was establishing media between these modes. Publishers would break down the barriers between them and their audiences, which could then enable them to become participants in the media. Conversational media was a natural progression to serve as an antidote to what John Battelle called “packed goods media.”

Most of that didn’t come to pass. The community-driven models of blogs showed early promise, only for those with escape velocity like TechCrunch and Mashable to morph into everyday publishers. The community turned out to mostly be confined to site comment sections. The internet’s inherent downside of fragmentation ended up sapping the growth of conversational media, enabling platforms like Facebook to become the de facto conversational backbone of the internet. We saw how that turned out.

Looking back, I don’t think the idea of conversational media was wrong. We’ve seen time and again that many ideas arrive at the wrong time and aren’t executed correctly. The promise of artificial intelligence leading to a search engine experience more like a natural conversation is not new. Early search forced people to use Boolean logic to talk robot vs simply ask for what they wanted. We needed to adapt to the tech instead of it adapting to us. Reversing that dynamic was the entire promise of Ask Jeeves and its “smart answers.” ChaCha promised something similar through a texting service. The problem is the answers weren’t that smart and required humans. AI has Silicon Valley aroused because it promises scale – and holds the promise of replacing expensive and inefficient humans in many processes.

Scaling conversations is hard. Technology does many things well, nuance is not one of them. Many of the downsides of social media – the bullying, the misinformation – are obvious in retrospect. The backlash building against AI Bing, which to be fair does seem to often become deranged, is a case of fooled me once, shame on me. The race to define AI moved more quickly than I thought from nitpicking over accuracy – this Mexico City itinerary includes a restaurant that doesn't look cozy – to dystopian fears of AI Bing Gone Wild.

There is plenty of material to choose from. At The New York Times, Kevin Roose has published his conversation with AI Bing, aka Sydney, in which the bot goes bonkers. There are similar examples from Digital Trends and Stratechery. The real gold in these conversations is when the robot can declare a desire to be human or to do harm. That’s when you know you have the good stuff. What this assures is that AI will become yet another culture war issue. Expect calls for regulation and litigation to come next. Imagine trying to figure out the obligations of ChatGPT and the others with regards to GDPR and the Right to be Forgotten. Then, on cue, the libertarian tech crowd will say this is just the “woke capitalists” at The New York Times and the establishment protecting their turf and power. New series, same general plot lines.

Publishers are at this point inured to external threats to their businesses. I imagine them reacting with the resigned “ah shit, here we go again” meme. With the promise of conversational media that established communities around publishing brands, the most successful bits of publishing companies have been in the blocking of tackling of digital media. Most publishers have SEO operations, shoveling service content onto the web to be crawled by Google and ranked. This is far more valuable traffic than the typical news content, for example. A news story has decaying value soon after it is published. It’s like buying a boat. But the service content produced to feed search engines sees its value rise and continue.

The biggest direct threat to publishers is this high-margin content gets squeezed. The lasting impact of the rise of AI search could be in how it will force change to the interface of search. The experience of using ChatGPT and even AI Bing is different from a typical search experience. For years, we’ve trained ourselves to prompt search engines in terms they would understand. A conversational interface changes all of that, training people to think conversationally. The big unknown is how normal people will use these tools. I have a hard time believing people want a robot that acts as “a kind of god that directs their actions.” Then again, the people peddling this are fresh off marketing stateless existence in the metaverse. Wait, what happened to the metaverse? Was that last season’s plot?

I don’t think how journalists, researchers and hackers are using these tools – trying to get a robot to deny the Holocaust or declare a secret desire to be human – are typical. But we’ve seen before where the assumptions of a new search technology turn out wrong. The smart speakers were meant to herald in an era of conversational search through speaking to Alexa. But most people use the Echo and similar smart speakers for basics like playing music and checking the weather. Technology tends to be overhyped and underhyped at the same time in its earliest phases – and the early assessments are nearly always wrong. I find it hard to believe that humans who are so distrustful of other humans will have more trust in a chaotic energy bot that sometimes flies off the handle. Maybe instead these tools will be less utility, more distraction.

The lasting impact of AI search could be a shift in behavior by making natural conversation a habit that extends to media, not just in navigation but in consumption. This was a topic we discussed in the forthcoming episode of People vs Algorithms. (Should be up later today. Sign up.) We’ve seen time and again how new technology platforms can shift expectations. Amazon set the expectation for fast, cheap (if not free) delivery. The search query being replaced by a conversational interface would fundamentally alter the most important interface in digital media. That could provide an opportunity for publishers who embrace a new form of conversational media. We’ll see gimmicky attempts to recreate a ChatGPT like interface – remember the Quartz chatbot? – but the lasting impact will be both in how content is produced but also the type of content. It’s not too far-fetched to expect publishers to have their own “branded’ AI that will enable conversational media trained on the brand’s point of view. A generic travel itinerary for Mexico City is less interesting than one produced from the point of view of a trusted person and/or brand.

For all the threats of these AI tools, I think more opportunities will arise for publishers to turn toward conversation as a core part of the media product.

One of the common themes of this newsletter is the need for publishers to have efficient infrastructures in order to have sustainable businesses. John Shankman, CEO of Hashtag Labs, an outsourced ad management firm for publishers, has seen the challenges in his experience as a publisher. I asked him his thoughts on common mistakes publishers make in tech:

The question here is really: Why do publishers feel compelled to build their own technology? One answer is that it adds enterprise value. The second and, I think more apropos answer, is that, generally speaking, the in-market tech solutions available to publishers don’t function exactly as the publisher needs them to. The “out-of-the-box” solutions aren’t configurable enough and don’t allow the publisher to give their core constituents (the audience, writers, editors, producers, and, yes, sometimes advertisers) exactly what they need. It really boils down to a fundamental question of build or buy. Are there solutions in the market that help a publisher achieve its goals better, faster and less expensively than doing it in-house? I know you’re only supposed to be able to get two out of three when it comes to good, fast and cheap, but it seems like the right tech solutions can really offer people all three.

Learn how HTL work with publishers to help them achieve their revenue goals.


Brian Wieser has long been one of the smartest media and ad industry analysts around. His new newsletter, Madison and Wall, dissects the industry’s fundamentals. He’s got a great breakdown of Trade Desk’s recent earnings that reads between the lines of its statements.

The Year of Efficiency continues in the tech industry. Airbnb managed to grow its revenue by 75% while cutting headcount by 5%. Cue the cleaning-fees jokes.

Meanwhile, Meta has shown signs it is weathering the storm of the challenges thrown its way by the rise of TikTok and Apple’s moves to kneecap Meta’s targeted ad business. CMO Alex Schultz is warning against complacency, even describing Meta as in the midst of a “turnaround.” I guess everything is relative: The company generated $32 billion in the fourth quarter.

Schultz signed off his message with “stay focused,” and the flight to focus means tech companies shedding some initiatives. Instagram’s abandoning its live shopping ambitions to focus instead on what it’s best at: selling ads. Live shopping has struck me as a trend that makes perfect sense for taking off – just look at how popular it is in China – but might never actually happen at a large scale. The same can be said of “super app” ambitions. Sometimes things that work in a market like China don’t translate to other markets with different dynamics.

The fate of Vice remains uncertain. The company raised another $30 million in debt financing to get itself some breathing room, although it won’t change the challenges facing the business as it figures out an M&A path during what, to be charitable, is a less-than-ideal time to be looking for a good valuation.

Newsletters are a tricky organizational fit at many publishers. The MediaVoices podcast has a good conversation with the Financial Times’ head of newsletters, Sarah Ebner. “You can’t do newsletters well unless you’re un-siloed and work across product, tech, marketing, commercial and everyone else.”

Thanks for reading. Send me a note with your feedback: You can also just hit reply to the email.

If you’re looking to reach 14,000 publishing professionals, you should consider sponsoring The Rebooting to showcase your expertise in finding paths forward for building sustainable media businesses. Check out the sales kit and get in touch.