Welcome to People vs Algorithms #83.
I look for patterns in media, business and culture. My POV is informed by 30 years of leadership in media and advertising businesses.
Sometimes it’s nice to read in the browser.
The AI Time Machine. PvA podcast out now. Apple | Spotify
Hello. It’s been a while.
We’ve had a few weeks of deep prognostication about the future of the web. This week on the PvA Podcast we tried to think things through a bit, asking ourselves “what does AI do to the web five years out?” Tune in.
If media types are feeling a little anxious, it’s understandable. Our attention space is deeply out of equilibrium. AI is a profound disruptor of the interface and creation layers of media, a giant new wave in a cascading tide of disruptions. Tight money has refactored priorities and pushed profit urgency. Cookie deprecation persists as an thorn in the side of the digital advertising ecosystem. UGC continues as a long term pressure point on media economics and hands huge advantage to the platforms that enable community creation. Streaming has brought on-demand economic forces to TV. And… search as navigational interface to the open web is about to go through radical change. This really has people’s heads spinning.
Managing through change in media is a game of synchronizing its three primary interconnected ingredients - distribution, content, monetization. Distribution is always the ballast. If it is in flux, everything goes haywire. Which is why the search conversation is top of mind for everyone in digital media. If you make money from selling attention or stuff in the digital world, it’s a challenging time to forecast the future. M&A is a nightmare.
It is not unreasonable to foresee a world of interconnected AI agents that completely annihilate the layers between your customer and an outcome. This may happen in some distant world. My bet the future will look more like the present than people think.
A couple of things to consider…
Google gets new competition
Everybody wants to control the AI chat interface because… duh… controlling the starting point of information discovery and commerce is ultimate power. The Ring.. Dark Lord Sauron, the Ring. What we are seeing is the proliferation of starting points. AI has created a competitive crack in Google's search armor.
“New Search” looks different than old. It sits at the intersection of the web index, AI chat and soon, smart agents that do things on your behalf.
You can see the New Search in the app Perplexity — it answers questions via the OpenAI API, it rolls these responses together with web results and positions the best sources as “citations” atop your conversation. It asks smart follow up questions to help you go deeper.
Google, Meta, OpenAI and Microsoft all liked these ideas too. They’ve caught up quickly. Today all offer a variation on the idea. New Search is quickly making its way to all of your interfaces.
Perplexity is a rounding error in the big picture. They have no material distribution. Which is why it is Meta’s moves to push a Perplexity-like experience to all of their touchpoints is worth noting. Now you can activate Meta’s new chat offering from a text box atop WhatsApp. And invite your friends to participate in the experience. They are doing the same thing on Facebook and Instagram. Billions of touchpoints.
ChatGPT offers the same intersection of AI chat and real time web search, but unlike Meta, they do not touch billions of people everyday. We will see the same on all Microsoft products, and soon as a core part of the Apple OS, potentially powered by OpenAI. At least until Apple can do it on their own.
Browsers are about to evolve too, shifting from dumb page rendering environments to AI-driven chat interfaces with knowledge tools enabled by AI. All browsers will integrate and expand the features of the New Search.
All told, Google with 90% plus market share has nowhere to go but down. The things you do on Google are gonna happen elsewhere. Google will fight back by taking attention share from the rest of the web, but will lose share as the place you start your information journey.
More distribution points mean more potential deals for media IP and more places to push content. More is better.
Google taketh.
Naturally, chat pulls knowledge and attention away from pages on the internet. This is what Liz Reid, Google’s head of search, means when she says charming things like “Google will do the Googling for you.” Which means it compresses user interaction on your start page - the search front end.
This is not a new phenomenon. Google has a long history of pulling utility into search, though they have had to do it judiciously lest they piss off the world and compromise the economic engine of selling links to the rest of the kingdom. Products, weather, flights, sports scores, movie times… all made the way to SERP, though integration of “general knowledge” is obviously a huge new battlefront.
So this sounds bad for anyone downstream. Indeed, much of the general, “evergreen” stuff people make to grind out positions on the web are no longer gonna trap traffic or make any money. You might license it to the LLMs but this will not make up the difference of search sourced traffic.
It will definitely start a new cycle of distribution soul searching to understand what you make now and how you optimize content output for a new reality. Fortunately, publishers have become used to the dance.
So, does chat replace all pages? Not yet. Not for a while.
Yes, some web pages are still useful.
At its best, a well crafted page — one that elegantly collects data, makes sense of it and packages it to answer — is still useful. Think about a page comparing hotel rooms or credit cards. There’s utility in the presentation layer. AI will do this better over time, but today a laborious back and forth with a chatbot is not better than a comparison page when you want an answer. Behavior will change more slowly than you think. Plus Google and others will have an economic incentive to make a place for these links.
Then there’s the media brand…
Media brands are useful too.
We are too dismissive about the organizational and cultural import of media brands.
Google is an information access and utility brand, not a point-of-view brand. I might trust their chat bot to tell me “who won the Peloponnesian War?” but not “which health insurance to buy?” or “dissect last night’s NBA playoff game with insight and wit.” The trust and reassurance of a media brand and the people behind it matters. It will continue to matter. Let’s not forecast out too far where your AI agent connects with the seller AI agent, evaluates all the options with perfect knowledge of your preferences and idiosyncrasies, agrees on a deal and celebrates over a digital martini. Slow down.
Inside of these new interface points, a trusted mark is a very valuable thing. I have seen this first hand in search. Brands that people trust have superior click through rates. The relationship between the people making the content and the brands that channel them may evolve, formats change, but the enduring long-term organizational functionality and trust of the best media brands will continue to matter even if how they are manifest in a new interfaces evolves.
As always, in an exhausting continual two decades of media upheaval, the best brands, backed by innovators, the technically adept and those that move the quickest will take share.
Search compresses.
As interaction moves to chat, link value will converge on a single top-performing result. This is sorta true today. The top few results garner most of the clicks on a search page. But when blue links turn to chat “citations,” it will compress further. Chat will reshape winning pages below the link. Winners will be data and functionally rich, informed human voices, differentiated POV.
Through this evolution, I suspect the big guys will crowd out many in the long tail. We are seeing this now. Outside of social platforms, it represents a return to media’s oligopolistic roots. Large well funded players will execute with more sophistication, their brands will resonate more with consumers inside of new interfaces. Plus they will matter more to Google and others cause they spend more for distribution.
Google cannot rely on single citations. They will have to rotate through sources to prevent civil war. Ten blue links become a “chat share of voice.”
Chat goes free. Just like search.
There’s a reason why OpenAI decided to make their latest release, ChatGPT4o broadly available for free. Meta has disrupted the space with free chat offerings funded with their huge ad spoils and the efficient open source LLAMA model. Apple will offer it as a free part of the OS. Microsoft has just demonstrated how more of the compute will move to your local machine with the launch of “Copilot + PC.” And AI chat is quickly becoming just part of how you search the web with Google.
Early thinking that AI chat would move to subscription is wishful thinking. Remind me to cancel my ChatGPT and Perplexity subscriptions.
All of which leads to the monetization conundrum…
Free chat means ads win.
While AI inference is becoming more efficient as all compute does over time, it’s still wildly expensive. Meta can pay for it with ads. So can Google. Both manage exceedingly complex advertising ecosystems with hard earned economic and interface connections to a long tail customer base. Apple has massive coffers, a services business and hardware profits to pay compute bills. Microsoft benefits from a mix of everything plus its B2B hegemony. All have massive distribution leverage, among other full stack advantages. But in the end, you need advertising to pay for things. Interface owners will make space for them.
Ultimately, OpenAI has to rely on an economic model dependent on selling AI services to B2B customers and owners of distribution (Apple). It’s going to be harder and harder for the company to maintain technology advantage (we are already seeing the feature gap shorten, see the OpenAI Spring vs Google I/O announcements), while paying for compute and growing a scaled sales and service operation. There’s gonna be a lot of roadkill in the AI space. OpenAI may be the next Netscape.
In short, an advertising business will become a necessity to pay for free consumer offerings. The ad side of these businesses are exceedingly resilient and painstaking for newcomers to replicate. Meta and Google are heavily advantaged.
Search optimization and paid distribution sustain.
Distribution changes always force change on the content creators. This time, SEO morphs to feed chatbots but the mechanics will be largely the same. Per above, all of the new chat distribution points will have to integrate featured ad spots in some shape or form. Same as it ever was. Google, Meta, Apple, Amazon will be the same gatekeepers. God willing, we get a couple of new ones.
Vessels for UGC do very well across board
AI chat aint gonna stop people from making content. In fact UGC will become increasingly important as grist for the machines. This is ground zero for knowledge. Platforms that host and harvest this content - Reddit, YouTube, TikTok, Insta - become an increasingly important media substrate… if that wasn’t already clear.
I am biased. I want a healthy media layer below the platforms. I want a media environment that sustains brands outside of social and outside of your email box. My experience suggests that the media ecosystem always finds ways to navigate distribution disruption. New cracks get filled, new opportunities emerge, content evolves. Time has also taught me that humans and businesses are way slower to change than we think.
But when change catches up, you best be ready. This shift is big. Best to start preparing now…/ Troy