ChatGPT is set to remain a hot topic for the rest of this year, a testament to its feverish appeal. Launched only last November, the AI chatbot grew to over 100m monthly active users in just over 2 months. According to Fortune Magazine, it is “the fastest-growing app in history.”
The chatbot, developed by Open AI, is an advanced language model that pools data from multiple domains to form intelligent, contextually relevant responses to queries in natural language. Its incredible potential has been met with a mix of excitement on one end, and concern on the other.
Right now, tech giants are in a race for AI dominance. Microsoft has invested heavily in Open AI and is moving to bring the capabilities we see in ChatGPT into their product stack, for an experience that would rival Google search. And Google, not to be outdone, recently announced their AI-enhanced chatbot, codenamed Bard. Among other advanced features, Bard will bring Google’s conversational intelligence into search.
The stage is set for a seismic shift in how we experience search, and AI is at the heart of it. But how do we parse this shift? To clarify what might result from making these advanced capabilities freely accessible, it helps to revisit the beginnings of search as we know it.
Getting Here
Since the internet took flight back in 1989, we’ve levelled the barriers to information access. This led us neatly into a period of information ubiquity – anyone could share anything at any time. I vaguely recall a time when many of us were finding our voices on WordPress and Blogspot, two of the more popular blogging platforms. Around the same time, we were finding each other on social media – rekindling old connections and making new ones. It wasn’t so long ago that we entered this brave new world.
That’s how the internet – not the library – became the preferred source of information. And all this information needed finding somehow. Enter Google, the leading search engine provider. Google launched in 1998, just in time for the dot com boom. Over the years, Google has become synonymous with inquiry and is – rather fittingly – viewed as the world’s information catalogue. Suffice it to say that Google’s dominance in the search space has never been threatened seriously.
But Google, and the current search model, are not perfect.
The Problem With Traditional Search
Search as we know it excels at presenting information that is already online. Think of it as a GPS for navigating the internet. We enter a search term, and Google points us in a few different directions, ranked by supposed relevance.
The results we get reflect the quality of the input. The model assumes we know what we’re searching for, well enough to represent it in a few characters. That is true most of the time, but not always. Sometimes, we don’t have a search term neatly lined up. Instead, we have real-world questions to answer. Here’s where things get sketchy.
We are increasingly bringing questions to Google. And these questions are complex and nuanced in ways that don’t map neatly to what’s already out there. That’s exactly where the current search model falls on its face – nuance and complexity.
An Intelligent Search Experience
I entered a search query into Google to source information on preventing burnout. The results featured what burnout is, how it differs from stress, the harm it causes, and how to manage it. These were all spread across the first 5 results. Then there was the occasional tip for preventing burnout, which was the question, to begin with.
Here’s what ChatGPT provided in response to the query, “How do I avoid burnout?”
In my opinion, this is a very focused list that is directly relevant to the question. It is also a better starting place for this inquiry than a list of 36.3m results sourced from the web in 0.45 seconds. Each point here represents a thought we could explore individually.
I further asked ChatGPT to develop point 3. Here’s what I got:
Admittedly this reads like a considered (even well-researched) response. The sentences are well constructed, and the logic is sound. Importantly, the AI’s ability to present a contextually relevant follow-up answer gives this chat dialogue the feel of an intelligent conversation.
A New, Improved Search Category
ChatGPT and AI bots represent a new search category, both conceptually and experientially.
The current search model, led by Google, does a great job cataloguing online information and giving it back to us when we search for it. We’ve had access to the world’s information for a while now. Oftentimes, however, our queries present questions and problems that require harnessing this information, not merely pointing to it.
This is where ChatGPT and AI bots excel over the current model. For the first time, we don’t merely have access to information, we can now wield it in ways that apply directly to real-world needs.
We’ve graduated from having a digital library at our fingertips to having an intelligent research assistant. That’s a big step forward. And we’re still in the honeymoon phase of this new technology, excitedly exploring its possibilities.
As we adjust to this shiny new tool, it’s worth noting that ChatGPT – and AI-enhanced search – are not a magic wand. There are imperfections.
Known Limitations of ChatGPT
The folks over at Open AI are clear on the limits of ChatGPT. Once logged into the platform, you’re greeted with a screen that features (among other things) these limitations:
- May occasionally generate incorrect information
- May occasionally produce harmful instructions or biased content
- Limited knowledge of world and events after 2021
The first two are especially concerning. Unlike humans, this tool does not quite know how to fact-check its answers. So while we embrace it as an intelligent research assistant, we should know that it’s not always as prudent. Numerous online reviewers have highlighted inaccuracies in the results.
That reminds us of the obvious – it is a search tool. However seemingly intelligent, ChatGPT does not excuse the need for intellectual agency in our approach to inquiry. We should not take the answers it provides at face value, but have some process for vetting and validating them.
Despite its limitations, ChatGPT has changed how we search for answers. This marks the not-so-humble beginnings of a new era.
A Brave New World
Microsoft is not shy about its intentions. In an interview with the Wall Street Journal about new chat features coming to Bing search, CEO Satya Nadella stated that users will be able to “have a full-on conversation around all the search data.” At the time of writing this article, Microsoft is also beta-testing their new Edge browser, said to have a sidebar with AI-enhanced features like summarising the current web page and writing an email in your preferred style.
It’ll be interesting to see how Google integrates Bard into their existing search experience. As it stands, Team Microsoft has what feels like a clear advantage in this new AI-enhanced search era. But Google search is native to all of us. And conversational intelligence is something that Google has already featured impressively well on their Pixel smartphone. The race is far from over.
Whatever happens, search as we know it is changing. That change is set to reinvent our relationship with information. The implications for the knowledge worker remain the subject of much debate and speculation.
Technology will do what technology always does: give us a level of agency we haven’t seen yet. What we do with this agency as employers, employees and everyday users, reflect who we are and what we stand for.
