Home | Blog | Contact

Possible GPT Side Effects

Much like the rest of the internet, I was completely blown away when I used ChatGPT for the first time back in December '22. Since then much has been written about the topic by a variety of news outlets and by other hackers across the internet. Topics like safety, bias, accuracy, impact to current jobs, and wild predictions have all been discussed ad nauseam. However, as the future of chat bots and generative text services emerge from companies like Microsoft and Google, I believe there is a topic that hasn't quite made it yet into the zeitgeist yet. What will be AI's side effect(s) as it is incorporated into search engines?

Search Engines in 1990s-2010's

As the internet gained traction in the '90s, there was an increasing need to make its content more accessible and organized. Companies like Yahoo! organized the content into directories; Ask Jeeves tried use natural language for a more natural Q&A experience; and of course Google's simplicity, speed, and quality eventually won out. Google's mission has been to "organize the world's data". Google's product was better in every conceiveable way and internet users flocked to them en masse. Users no longer "searched" the web, they "googled" it.

Google then leveraged its dominant position to take advantage of its users. It was able to capitalize on user data in the form of online advertising, it maintained its dominant "default" position through exclusivity deals with browser and other software vendors, and in-turn the profits it reaped it reinvested heavily as a part of its virtuous flywheel. Its early investments allowed it to catapult into the stratrosphere of internet history by becoming an immensely profitable internet juggernaut, a position it still enjoys to this day. However, this isn't a post about a retrospective on Google, but rather search engines.

In the time between the popularization of search engines through the early 2010's, the relationship between users, the search engine, and website owners remained largely unchanged, which was:

  1. Users visited their favorite search engine to perform a search,
  2. Search engines would return relevant links,
  3. Users would click a link from the search results,
  4. Website owners would get a visit from the user

This relationship was about to shift as search engines became even more powerful and could organize the world's data better.

Search Engines in 2010s-2020s

In 2012 Google introduced the world to Knowledge Graph in a blog post which started with the words, "Search is a lot about discovery.." They touted how it would allow users to find the information they were looking for directly within Google's search queries without having to clickthrough to the original source. It operated on well structured sources like Wikipedia and CIA's World Factbook, but its impact more far reaching. It influenced a generation of search engine developers to no longer act merely a waypoint for user's internet search journey, but as its terminus.

No longer did users have to rely on search engines to find and organize relevant links, but they could rely on search engines to also present information directly to them. Website owners like Yelp, Amazon, Expedia, and others, as well as regulators like US DoJ, have long recognized these consequences of search engines no longer became content with being waypoints. The new paradigm shifted the relationship to:

  1. Users visited their favorite search engine to perform a search,
  2. Search engines would return relevant information and links,
  3. Users would either be content with the information returned, or they may click through a link from the search results,
  4. Website owners would enjoy some clickthrough.

To ensure that users discovered what they were looking for, search engine companies continued to iterate on step #2 to reduce the need to clickthrough off of their sites. However, nobody stopped to talk about what would happen to website owners if they became wholly redundant.

Search Engines in 2020s and beyond

Microsoft's Bing and Alphabet's Google currently enjoy a ~95% market share for search engine traffic in the US as of February 2023 according to StatCounter.com. Both of these companies have both announced in early 2023 their intentions to include a ChatGPT-like experience to their search engines. If they are successful at conquering the challenges that lay ahead of them - i.e., accuracy, bias, safety, etc. - then I would posit that this shift will impact every single website owner on the internet today.

For some time there will be categories of search where users will need to continue to click through until the AI results are good enough, but I cannot think of a single category that couldn't be ultimately organized and centralized into this new paradigm. I think we'll quickly see the paradigm of search engines shift to the following:

  1. Users visited their favorite search engine to perform a search,
  2. Search engines return relevant information
  3. ...done?

As far as I know, nothing can stop this inevitability. The current search engine giants are going to use AI to collectively organize the world's information. They will use this information to become the terminus of knowledge for users. They will no longer act as gatekeepers that point or bias their users to a set of web properties that match their algorithm.

I hope I am wrong, but I see this another step to the centralization of the internet, and the ossification of the current internet giants. Perhaps Bing will finally be able to pass Google as they were asleep at the wheel, and time will tell if that happens, but things are about to get so much worse for small-to-medium website owners.

Either way, you're probably not reading this on my website anyways. Our AI overlords have in all likelihood summarized this article into a more manageable format for your consumption. Long live our AI overlords.