Suchmaschinen-Chatbots sind gekommen um zu bleiben - das musst du für deine Website wissen
Microsoft announces ChatGPT integration with Bing

Search-Engine chatbots are here to stay – here's what you need to know for your website

Last week, Microsoft debuted its integration of ChatGPT in the Bing search engine. The implementation of ChatGPT is no big surprise as Microsoft is one of the early-stage investors of OpenAI, the company developing ChatGPT.

Johannes Terhürne
Johannes TerhürneFounder, Managing Director

In the demo, Microsoft showed how AI-powered chatbots could improve the search experience for users. Next to the regular search results, Bing now shows a summary of the search term. This summary can contain advanced search queries like vacation planning. It is important to emphasize that the results rely on current information. As ChatGPT in the OpenAI demo uses knowledge of late 2021, the implementation in Bing also considers new information. Microsoft does not provide which version of the GPT language model it operates in the Bing search bot. Still, we can assume that Microsoft is using an optimized version of the GPT-3 model or even a very early version of the new GPT-4 model that will launch later this year.

The summary contains information from different sources. Microsoft attributes the sources directly in the text, and users can visit these sources to get more insights.

In addition to the new search experience, Microsoft also introduces ChatGPT in its Edge browser. Users get an additional toolbar that allows users to search inside the current tab for information. This feature can help users to search inside a page. For example, if a website shows the earning reports of a company, a user can ask for the revenue of a specific month, and the chatbot will search for the exact figures and display them to the user. The chatbot is also able to write summaries, so users can ask the chatbot to write a short summary about the content of a page or document.

These new features are a good indicator of how the future of the internet can look like. Google has also announced that it will add an improved AI-powered experience to its search engine this year. This is just the newest development in the context of conversational user interfaces. Since the first leaps into this sector with voice assistants like Amazon Alexa or Apple Siri, which are very limited in the information they can provide, OpenAI showcases its newest development in the sector of artificial intelligence which marks a new milestone. Chatbots are now able to access almost limitless information sources and aggregate this information for users. What sounded like science fiction only ten years ago is now about to become the new normal for millions of users worldwide.

But with this development, new challenges arise, as well for companies and brands as for legislators and state actors.

For legislators and state actors, a lot of legal questions are becoming important: When an artificial intelligence aggregates different sources, how does it take care to filter out false or even fake information and tells them apart from facts? For brands, a real-world scenario could be that opponents are feeding fake news about the brand into an AI. With no filters, this information would spread and could become a new way of character assassination. This can also become very dangerous even for societies, as the same mechanic could be used by terrorists or regimes to destabilize markets and populations.

In ancient Rome, satirist Juvenal asked sarcastically, "Quis custodiet ipsos custodes?" which means "Who will guard the guards"? The European Union approved the Digitial Services Act in April 2022, which targets companies that build digital services that could be the root of spreading misinformation. Tech companies like Google, Facebook, Twitter, and Apple will need to provide further information on how their algorithms are working and how they are fighting misinformation in their networks. For this, the European Union has also founded the European Centre for Algorithmic Transparency which will control big tech companies and make sure that all European laws for data analysis and artificial intelligence are followed by the companies.

Aside from the legal perspective, there are also a lot of things companies and brands need to consider now. Since the beginning of the internet, it was always important to take care of search engine optimization (SEO). Most traffic for websites still comes from search engines like Google or Bing. 

This paradigm is about to change with the success of conversational user experiences like chatbots. Right now, a user necessarily needs to visit a website to get all information. In the future, this will not be the case, as conversational user experiences are based on presenting all relevant information and any context directly. That means that brands and companies will have fewer users visiting their websites as most information is already presented by the chatbot. This also means that brands and companies cannot control the narrative of their brand to the extent they can control it now.

The big question is how brands and companies can take advantage of this development. In our view, websites will still exist in this new normal, but conversational experiences will be the main lead source for websites. Therefore, it is important to store the data in an accessible way. Monolithic content management systems like WordPress, Typo3, and Drupal store information in a page-based manner. In this paradigm, everything is displayed as a webpage. Often the content is even stored as pre-rendered HTML inside of the content management system. Of course, monolithic systems can be extended with plugins that can store the data in a more object-based manner but in the default setup, most monolithic content management systems store content in a pre-rendered way. This makes it hard to recompile the information for other services that do not work on a page-based approach.

This is where headless content management systems like Storyblok or Contentful become important. Headless content management systems allow developers to define a data model for specific data types and provide the data via an API. Here is a quick example to make this approach a bit more understandable: 

We want to build a huge recipe platform. Our platform mostly contains recipes for different dishes. Each recipe has a list of ingredients, measurements for the ingredients, cooking steps, and kitchen tools that are used for cooking this dish. All of these different data types are stored inside a custom model called "recipe." If I want to access all the ingredients, I can address these directly through the API without loading all other information. In a monolithic system, this information would have been rendered as long HTML text. In a headless content management system, the information is still uncompiled and can be combined, altered and formatted by the requesting party itself.

This API is fed into different frontends. For a webpage, frameworks like NextJS or GatsbyJS can generate regular websites from the provided data. The API can also be utilized for mobile apps but also as a data source for other data services like chatbots. 

This is the most likely way how data will be fed into conversational user experiences like ChatGPT. Marketing teams need a new skill to determine which information needs to be fed into artificial intelligence systems. Next to the role of search engine optimization (SEO), a new role for artificial intelligence optimization (AIO) will emerge. AIO will decide which information will be provided to artificial intelligence. It will also be important to monitor the presence of the information in these systems and keep track of the reporting of the different platforms. AIO teams will also be responsible for tracking if misinformation is shared about a company and its products and file reports to the platforms to remove misinformation.

All in all, we have an exciting future ahead. The internet is about to experience the next revolution in how users access information. Still, there is a lot to do to make sure that information is based on core democratic rules and that misinformation cannot spread easily. For brands and companies, a whole new skill set is needed to make sure that their content is still visible on the search engines of the future. Therefore, thinking about a holistic and object-based content management system is key for future success. It is also important to prepare marketing teams for the challenges of tomorrow and create strategies to make sure that the company's content is still relevant to the new interactive experiences and users get valuable and important information firsthand.