Welcome to the forefront of conversational AI as we explore the fascinating world of AI chatbots in our dedicated blog series. Discover the latest advancements, applications, and strategies that propel the evolution of chatbot technology. From enhancing customer interactions to streamlining business processes, these articles delve into the innovative ways artificial intelligence is shaping the landscape of automated conversational agents. Whether you’re a business owner, developer, or simply intrigued by the future of interactive technology, join us on this journey to unravel the transformative power and endless possibilities of AI chatbots.
AI chatbots are strong at answering common questions and complex queries, but Google still has an advantage with information requiring absolute accuracy, breaking news, and high-risk areas.
AI -powered searches yield better results for common questions or complex requests that require aggregating information from multiple sources. Conversely, traditional search engines like Google are more suitable when users need high accuracy, want to choose their own sources, or are looking up sensitive issues such as health, law, or finance.
Claude, ChatGPT, and Gemini can all search for information on the web, in addition to many other functions. This may be one of the reasons Google is actively promoting AI Mode on its search engine.
When asked to search the web, the latest AI chatbot versions combine two sources of information: results from traditional search engines and knowledge extracted from the data used to train the AI. Everything is then synthesized and presented in the familiar style of generative AI, with polished, clear answers.
The difference lies in the additional AI elements beyond the usual search process. Therefore, AI-powered search isn’t always the best option.
For simple and common information, such as the height of the Eiffel Tower (France), AI-powered searches are a relatively safe option. Such data is not difficult to find on the internet.
Large-scale language models, the technological foundation of AI chatbots, tend to predict and select information that is highly likely to appear. Therefore, the likelihood of a chatbot providing an accurate answer in this case is very high.
AI is also useful for complex questions that traditional Google struggles to fully address. For example: “I want to find the best restaurants in Ho Chi Minh City for a family of 5, with a total cost not exceeding 4 million VND.” With this request, AI can synthesize information from multiple sources and provide a refined answer. Meanwhile, traditional search engines might only display a list of links sorted by more basic criteria.
A notable conclusion from the Stanford University study is that when the correct source is found, AI is very good at extracting relevant data from that source. The weakness often lies in finding reliable sources, due to the lack of suitable sources on the Internet or the AI misunderstanding the question. Even so, chatbots can still respond with a very confident tone.
Therefore, the less common the question, the more thorough the verification needed. For example, the name of the fifth-ranked actor on the cast list of a little-known television show might only exist in one or two sources. In that case, the AI could easily choose the wrong source and give an incorrect answer.
Similarly, the chatbots have never been to Ho Chi Minh City or dined at those restaurants. They rely on reviews and public sharing from people who have experienced them. When no website matches perfectly, the AI will search for content with related terms such as “affordable” or “family-friendly” and generate a response from there. For less popular locations or overly specific requests, such as “round tables,” the results may be less accurate.
AI chatbots are built to rephrase, synthesize, and summarize information. These processes, however, can also lead to errors. If you need highly precise details, such as exact quotes from celebrities, data in reports, or equipment specifications, traditional Google searches are often better.
This method is also suitable when you want to decide your own information sources. Delegating the search to AI is convenient, but it means relying on the chatbot’s ability to correctly understand the question, find the right source, and extract accurate information. This method saves time, but users have little control over the entire process.
Researchers at the University of California – San Diego (USA) discovered that AI sometimes prioritizes sources with content closely matching the user’s question over those with higher credibility. This means a personal blog might be preferred over a reputable medical website if the chatbot believes the blog is more relevant to the user’s search. With traditional search results, users are less likely to make such a choice because they usually prioritize sources they trust the most.
If you still need to use AI for sensitive information, carefully check the source the chatbot uses, even comparing it multiple times. AI chatbots often include source links directly in their responses. If you’re unsure, you can ask the AI to explain where the information came from to determine if it’s content generated by a large language model.
Of course, traditional Google searches can also lead to misinformation. However, without chatbots, you’re in a better position to assess for yourself what information is accurate and what is unreliable.
Overall, provided there is regular verification, AI-powered search is most effective for common questions that are more likely to find accurate and up-to-date sources. However, for questions where accuracy is more important than convenience and speed, traditional search methods still have their place.
DEPARTMENT OF GRASSROOT INFORMATION AND FOREIGN INFORMATION
MINISTRY OF CULTURE, SPORTS AND TOURISM
Responsible for content
Mr. Pham Anh Tuan
Head office
