Launched in November 2022, ChatGPT swiftly emerged as one of the most popular websites in the world. Millions of people have used it for all kinds of purposes ranging from writing computer code and bug fixes, to generating entire children’s stories. With the recent introduction of GPT Plugins such as Browsing, people are increasingly using ChatGPT to find answers that they would normally find via Google such as answers on Stack Overflow.
Does this mean ChatGPT will eventually replace Google as the dominant search method on the internet? Should Google be worried that their long-standing reign as king of search might soon be challenged?
How ChatGPT has grown to compete with Google
When ChatGPT was first released, it was not a Google killer, as it lacked a connection to the internet and therefore could not act as a fully-fledged, up-to-date search engine. However, Microsoft has taken significant strides to enhance ChatGPT’s capabilities by introducing Plugins and integrating ChatGPT with Bing search.
In 2019 and 2021 Microsoft invested money and Azure computing resources in OpenAI. At those times, OpenAI was still developing GPT-2 and GPT-3 without clear ideas of where they would lead or how they might impact society. Recently, Microsoft announced a third investment in OpenAI. This time, however, they have clearer intentions of integrating ChatGPT-like models with Microsoft services and products, such as the integration of ChatGPT with Bing search. Microsoft asserts that the new Bing search experience is the “culmination of four technical breakthroughs'' including:
- Next-generation OpenAI models built on advancements from ChatGPT and GPT-3.5
- A proprietary way of working with OpenAI models that allows Microsoft to leverage their power
- The application of AI to the core search algorithm to make the biggest improvement to accuracy and relevance in two decades
- A new user experience that changes how users interact with search, browser, and chat
The updated Bing is accessible to anyone using the Microsoft Edge browser and includes a chat interface positioned alongside more traditional search results as well as a pure chat interface similar to ChatGPT. This enhanced Bing experience may be a foreshadowing of how everyone will interact with search/the internet in the near future. And this may be how Microsoft finally unseats Google as king of search.
Google has possessed a conversational model called “LaMDA: Language Models for Dialog Applications” for several years but has chosen not to release it likely due to concerns about jeopardizing their primary revenue stream from search. Since Google’s existing product is successful, they may be hesitant to introduce a chat AI that could potentially provide false answers.
In contrast, OpenAI had nothing to lose by releasing ChatGPT. Moreover, Microsoft, trailing behind in the search market, can afford to take greater risks in the pursuit of innovation. Users might be more forgiving of Bing providing incorrect information, attributing it to Microsoft's ongoing development efforts. OpenAI and Microsoft’s actions, however, are forcing Google to react, so much so that they issued a “code red” in response to ChatGPT’s popularity, eventually leading to the botched release of Google’s own conversational AI service called BARD powered by LaMDA in February. However, their caution is still evident as they continue to maintain a limited waitlist restricting access and include a disclaimer about the experimental nature of the service, warning users of possible inaccuracies or inappropriate responses.
Is AI-infused search here to stay?
The debate remains whether chat AI-infused search is a game-changing innovation or merely a passing trend. Tom Scott, a popular tech YouTuber, believes it is signaling a change in technology similar to when the internet was first becoming popular. Meta’s Chief AI Scientist Yann LaCun, on the other hand, has said that ChatGPT is 'not particularly innovative,' and 'nothing revolutionary.' He continues: "It's nothing revolutionary, although that's the way it's perceived in the public," said LeCun. "It's just that, you know, it's well put together, it's nicely done."
LeCun may be overlooking the critical aspect that ChatGPT's well-executed design is what makes it impactful. The success of future applications of ChatGPT and other Large Language Models (LLMs) will depend not only on their technological advancements but also on how effectively they are integrated with other systems and made accessible to users. This distinction often determines whether a data science or software project will create a meaningful impact or be shelved and forgotten. A well-structured project is crucial for achieving significant outcomes within an organization.
This also helps explain why Google's LaMDA, which may be technologically equivalent or superior to ChatGPT, has not made a substantial impact on its business. Meanwhile, Microsoft is rapidly working to leverage ChatGPT in an attempt to outpace the search giant. LeCun's perspective misses the point: innovations that change the world are often not significantly more advanced on a technical level; rather, they are effectively packaged and presented in a manner that maximizes their impact beyond their inherent qualities.
In conclusion, ChatGPT's rise highlights the importance of not only developing cutting-edge technology but also ensuring its seamless integration with existing systems and user-friendly access. As Microsoft and OpenAI continue to push boundaries with ChatGPT, it remains to be seen whether this innovation will ultimately disrupt Google's long-standing position as the king of search.
This post is part of a series about the explosion of AI happening right now. Check out the first post about the timeline of OpenAI's model development and our second post on the acceleration of large language models like ChatGPT.
Photo credit: Daniel CHETRONI - stock.adobe.com