How Will AI Chatbots Revolutionize Search Engines?

Ashley_Buckwell

Moves by search providers to integrate AI chatbot technologies into their search engines will change how we find information online. An AI search engine would bring big advantages, but it also has the potential for some negative unintended consequences. In this article, we dive into the effects AI chatbot technologies are likely to have on search engines of the future. 

What is an AI chatbot?

As we wrote about last month, AI chatbots are the latest iteration in chatbot technology. These tools take advantage of recent developments in the field of AI to provide personalized, contextual responses to user queries, and they can even provide accurate answers to complex questions, although not always.

AI chatbots are built on large language models: algorithms trained on vast chunks of the Internet. An AI chatbot predicts the likeliest next word in a response to some query based on its reading of billions of sentences that use the preceding words.

The most well-known AI chatbot today is ChatGPT by OpenAI. The language learning model behind the chatbot, GPT-3, is a combination of machine learning algorithms and enormous training data sets that allow the chatbot to function. GPT-3 is the third generation of GPT technology and was the most powerful language model when ChatGPT was released, but upgraded versions of GPT are already in the works that will provide even better results.

Some of the tasks ChatGPT can perform include the following:

• Answer questions (both simple and complex) 
• Write computer code 
• Write articles, poems, stories, genuine-sounding tweets, and much more 
• Respond using the writing style of a specific person 
• Summarize long documents 
• Solve math problems

Several other companies, including Google, IBM, and Meta, are currently in the early stages of developing their own AI chatbots, but these are either still in testing phases or have strictly limited availability. Furthermore, ChatGPT’s functionality is considered to be the best as of now. 

Integration with search engines

Three of the biggest search engines announced this year their intention to integrate AI chatbots into their search products. Google, Bing, and Baidu are all currently competing on this front. Furthermore, several smaller search providers, including Neeva and Kagi, have either announced plans or begun to incorporate versions of the GPT language learning model into their search products. These developments will allow users to get direct answers and engage in conversations, rather than merely receive a list of links after typing in a word or question.

In early February, Microsoft launched a new version of their search engine, the Bing search engine, powered by a combination of GPT and Microsoft’s own Prometheus language learning model. Access to the new Bing is limited by a waitlist, but Microsoft says it will open up the experience to millions of users in the coming weeks.

Here is a summary of the experience offered so far by the new Bing. The first change users will notice is that the search box invites users to ask it anything. If you want to keep using keywords, it will happily use those, but you will get the best results when you ask it a more open-ended question. The next change is that the returned search results include a combination of old-school links plus new AI featured summaries, which include citations to sources of information. When you ask it for something specific, it will give you the AI-powered results at the top of the search results page, and for longer, more complex searches, it will provide AI-powered results in the Edge browser sidebar with chat options to delve deeper into the results. The AI functionality can be inconsistent, as Bing will sometimes seemingly forget that this new experience even exists and will only return traditional links.

Google has not yet integrated AI chatbot technology into Google Search, although it does intend to eventually. As of writing, Google has opened up its experimental conversational AI service, Bard, to trusted testers, and will be making it more widely available to the public in the coming weeks. Eventually, Google plans to add AI-powered features to Google Search that will enable it to distill complex information and multiple perspectives into easy-to-digest formats. 

Kagi is another search provider that has been actively experimenting with AI. The company uses GPT in the backend to provide semantic search functionality, allowing it to understand natural language queries and to eventually be able to provide more conversational results. In January, Kagi offered a preview of its universal summarizer, an AI tool that summarizes any content on the web, given just a URL. 

How will AI chatbots change the future of search engines?

The initial deployment of AI chatbots as either add-ons to search or as standalone conversation partners makes sense, given their occasional inaccuracies. However, as their capabilities improve, chatbots are going to either compete with search engines or complement them.

Here are some of the ways AI chatbots will change search products in the future:

• Search will become conversational. Instead of serving up a list of links, chatbots can return fluent answers. 
• We will get instant answers to complex questions. Rather than scrolling through multiple links from multiple search queries, a user will be able to ask one complex question with multiple variables and get a response that is distilled from the entire dataset of the search engine. 
• Online shopping will become easier. A chatbot can be asked for its top product recommendation in a specific category based on the parameters you give it, such as price, dimensions, and other features.

Furthermore, the search market is likely to expand in exciting and disruptive new ways. For example, software engineers are already asking ChatGPT to write code snippets. AI chatbot–powered search applications might include giving a diagnosis of an illness, summarizing a large text or website, and recommending the best flight ticket based on the user’s preferences. Companies may integrate AI chatbot technology into their in-house data search tools.

Limitations and ethical concerns

Chatbots are known to in some cases provide inaccurate information. For example, when an Insider journalist tested the model by prompting it to write a news story about a jeep factory in Illinois, the model’s output included some convincing yet completely made-up Tweets supposedly made by the factory’s CEO. Consequently, users must interpret responses cautiously.

Chatbots have occasional biases and prejudices as they scan the Internet. For this reason, it must be clear where information comes from, especially if an AI is drawing pieces from more than one source. Bing’s ability to cite sources within its AI-powered responses is a step in the right direction here. To account for bias, the Neeva search engine allows users to control and see what is happening in the news across the political spectrum by using a slider, and tools like this may be used in future AI-powered search.

Advertising in search results is likely to be affected. Advertisers may be charged more for the ability to influence the answers chatbots provide or to have links to their websites embedded in responses. If you ask ChatGPT today to recommend you a car, it will reply that there are lots of good brands, and it depends on your needs. Future chatbots may be more willing to make a recommendation.

Chatbots have many challenges to overcome, but they also offer valuable opportunities to those who wish to make online information more accessible. The race has just begun. We will be looking to see which AI-powered search provider becomes the new front door to the Internet.