Discoverability Crisis: The Marketing Storm On The Horizon
How in the hell is content discovered if the answers to the public’s questions are provided instantaneously without any citations?
Like many of you, I’ve been spending the past several months diving deep on generative artifical intelligence, how it works, how I can make use of it, and how it will reshape strategic communications writ large.
This post expresses thoughts about:
Discoverability in a citationless world
The paths of generative AI development
The flood of AI content
Preparing for our AI future
Brand Discoverability In An AI Chatbot World
My reaction to trying ChatGPT the first time was: How in the hell do organizations or content creators get discovered if the answers to the public’s questions are provided instantaneously without any citations?
I’ve been optimizing content for my entire career. It’s clear to me that this massive shift that is underway as a result of generative AI is going to require a fundamental rethinking of how marketers optimize for discovery.
At the moment, however, we don’t have much to fear about losing the ability to connect with audiences through search but that will quickly change.
Bing Chat vs ChatGPT vs Bard: Testing News Discovery
I searched for the lastest news about AI using ChatGPT, Bard, and Bing Chat. These are the results.
Bing Chat delivered by far the best results.
It supplied me with links to articles, some of which were just an hour old. Conversely, it also gave me two stories from the New York Post that appeared to be dated, and the link Bing Chat provided for those articles was to the index page for the Post’s articles that had been tagged with “artificial intelligence.”
ChatGPT summarized one USA Today article about tech luminaries signing an open letter warning of human extinction from AI. Every citation was a link to that same story.
The one company you’d expect to ace this test with absolutely no sweat at all—the one company that has been perfecting search results for news since the advent of search engines—failed miserably. These are the results from Bard:
Let me count the ways Bard failed:
OpenAI has NOT released GPT 4.5
IBM has been training Watson in an attempt to help diagnose cancer but to date, it has made no claims that it can, as the headline Bard gave me confidently asserted. But this one’s mostly accurate.
I guess that last one may be up for debate but I did a Google search using quotes around the headline to force an exact query and the search engine did not return any articles with that headline. The description cites “A new study,” so presumably there would be an article about it.
Of the top three, it looks like Google’s Bard “hallucinates” the most. When I asked for sources and URLs, Bard told me it wasn’t programmed to assist with that.
In other words: You’re on your own, bud.
But don’t let the poor results fool you. These AI chatbots are going to get better (as you’ll see with Google’s Generative Search Experience example below).
The ease and quickness with which users get their answers will train consumers’ expectations of how they interact with content. This is the discovery storm clouds that are already forming along the horizon.
The Reputation Algorithm is a reader-supported publication. If you want the really good stuff, you’ll need to become a paying subscriber
Keep reading with a 7-day free trial
Subscribe to The Reputation Algorithm to keep reading this post and get 7 days of free access to the full post archives.