Search And ChatGPT: You Still Can't Get There From Here

I’m wrapping up my ChatGPTrilogy with a shout-out to an old friend who will be familiar to many MediaPosters: Aaron Goldman. Thirteen years ago, Aaron wrote a book called "Everything I Know About Marketing I Learned from Google."  Just a few weeks ago, in a postentitled “In a World of AI, is Everything I Know about Marketing (still) Learned from Google?,” he discussed the last chapter of that book, “Future-Proofing.” Part of that chapter was based on a conversation Aaron and I had back in 2010 about what search might look like in the future.

Did we get it right? Well, remarkably, we got a lot more right than we got wrong, especially with the advent of natural language tools like ChatGPT and virtual assistants like Siri.

We talked a lot about something I called “app-sistants.” I explained then, “the idea of search as a destination is an idea whose days are numbered. The important thing won’t be search. It will be the platform and the apps that run on it. The next big thing will be the ability to seamlessly find just the right app for your intent and utilize it immediately.”



To be honest, this evolution in search has taken a lot longer than I thought back then.

Searching for complex answers as it currently sits requires a lot of heavy lifting. In that discussion, I used the example of planning a trip.  “Imagine if there were an app that could keep my master intent in mind for the entire process. It would know what my end goal was, would be tailored to understand my personal preferences and would use search to go out and gather the required information. When we look at alignment of intent, [a shift from search to apps is] a really intriguing concept for marketers to consider.”

So, the big question is, do we have such a tool? Is it ChatGPT? I decided to give it a try. After feeding ChatGPT a couple of carefully crafted prompts about a trip I’d like to take to Eastern Europe someday, I decided the answer is no. We’re not quite there yet. But we’re closer.

After a couple of iterations, ChatGPT did a credible job of assembling a potential itinerary of a trip to Croatia and Slovenia. It even made me away of some options I hadn’t run across in my previous research. But it left me hanging well short of the “app-ssistant” I was dreaming of in 2010. Essentially, I got a suggestion -- but all the detail work to put it into an actual trip still required me to do hundreds of searches in various places.

The problem with ChatGPT is that it gets stuck between the millions of functionality siloes -- or “walled gardens” -- that make up the Internet. Those “walled gardens” exist because they represent opportunities for monetization. In order for an app-ssistant to be able to multitask and make our lives easier, we need a virtual “commonage” that gets rid of some of these walls.

This conflict between capitalism and the concept of a commonage goes back at least to the Magna Carta. As England’s economy transitioned from feudalism to capitalism, enclosure saw the building of fences and the wiping out of lands held as a commonage.

This history, which played out over hundreds of years, was compressed into a few decades online. We went from the naïve idealism of a “free for all” internet in the early days to the balkanized patchwork of monetization siloes that currently make up the Web.

To go back to the idea of trip planning, I might be able to see what it costs to fly to Rome or the cost of accommodations in Venice is on a search engine, but I can’t book a flight or reserve a room. To do that, I have to visit an online booking site. If I’m on a search engine, I can manually navigate this transition fairly easily. But the process would stop something like ChatGPT in its tracks.

When I talked to Aaron 13 years ago, I envisioned search becoming a platform that lived underneath apps that could provide more functionality to the user. But I also was skeptical about Google’s willingness to do this, as I stated in a later post where I hypothesized that this might be an easier transition for Microsoft to make.

Whether it was prescience or just dumb luck, it is indeed Microsoft taking the first steps toward integrating search with ChatGPT, through its recent integration with Bing. Expedia (which also has Microsoft DNA in its genome) has also taken a shot at integrating ChatGPT in a natural language chat interface.

This flips my original forecast on its head. Rather than the data becoming common ground, it’s the chat interface that’s popping up everywhere. Rather than tearing down the walls that divide the online landscape, ChatGPT is being tacked up as window decoration on those walls.

I did try planning that same trip on both Bing and Expedia. Alas, Bing also left me well short of my imagined destination. Expedia -- being a monetization site to begin with -- got me a little closer, but it still didn’t seem able to get me to where I wanted to go.

I’m sorry to say search didn’t come nearly as far as I hoped it would 13 years ago. Even with ChatGPT thumbtacked onto the interface, we’re just not there yet.

1 comment about "Search And ChatGPT: You Still Can't Get There From Here".
Check to receive email when comments are posted.
  1. Aaron Goldman from Mediaocean, April 19, 2023 at 7:23 a.m.

    Always fun looking ahead and behind with you, Gord. To your point, we're not there yet but getting closer indeed. I think plugins will change the game as these roll out with GPT4:

Next story loading loading..