TL;DR
Explore Full Report
Interviewees spoke at length about the experience of using their chatbots of choice. For the most part, they describe a positive and pleasant experience. Watching them interact with the chatbots provided a more detailed understanding of their routines.
How We Did This
Using the Respondent research platform, CNTI recruited adults who said they (1) use AI chatbots at least once a week and (2) “keep informed about issues and events of the day” at least somewhat closely. To learn about the breadth of use cases and opinions, we sought to maximize variation across demographics. See topline for details.
Our interview protocol incorporated a concurrent thinkaloud approach. After a series of questions about general news and information habits, we asked interviewees to share their screen while demonstrating how they use one or more AI chatbots of their choice. We also asked them to walk us through AI chatbot interactions from their history and their use of other platforms and tools, including news aggregators, social media and news sites. These methods provide richness and depth; however, it’s not possible to generalize about the frequency of behaviors from these interactions, so we have refrained from using quantitative terms throughout this report.
CNTI’s analysis focused on reasons for seeking information identified by audience practitioners, the experience of interacting with AI chatbots and interviewees’ broader understanding of information.
The AI chat window terminal is a deeply personal space and the researchers in this project are incredibly grateful to interviewees who opened this safe space to them.
As with all CNTI research, this report was prepared by the research and professional staff of CNTI. This project was financially supported by CNTI’s funders.
See “About this study” for more details.
🇺🇲🇮🇳 Interviewees in both countries find the clean and structured presentation of information easy to process and enjoyable to read.
Most interviewees like that AI chatbots present information in a hierarchical manner that is easy to scan. Interviewees consistently said they prefer this structured presentation over the cluttered interfaces of traditional media and search engines. They pointed to specific features: short bullet points grouped under clear headlines with strategic bolding and a lot of white space. For many, the value lies in the AI chatbot’s ability to strip away noise (e.g., advertisements, sidebars, paywalls, irrelevant links) and present a synthesized hierarchy of information. In fact, the layout of the answer appears to be as important to interviewees as the content. All in all, it’s “really easy on the eyes,” said one U.S. interviewee. One Indian interviewee shared that AI chatbots like Perplexity had replaced web search for them because of their “clean UI” and the ability to get a direct summary without navigating through multiple newsletters or paywalled articles, saying, “earlier, if I wished to search anything I would go to Chrome but nowadays I either go to Gemini Live or to Perplexity most of the times, instead of the browsers.”
🇺🇲🇮🇳 Interviewees experience AI chatbots as a faster and more efficient way to get information than either search or browsing.
Many interviewees in both countries noted AI chatbots’ ability to save time. Compared to traditional search engines, some interviewees said querying an AI chatbot is more efficient and less overwhelming, allowing them to avoid clicking through lots of links and save time on a variety of tasks. They particularly value saving time on tasks they do not enjoy or do not find valuable, which range from formatting PowerPoint presentations to looking up tedious information. One interviewee from India utilized ChatGPT to plan a detailed 5-day travel itinerary for Dubai. They noted that doing it manually through travel blogs and booking sites would take “four to five hours,” whereas the AI chatbot came up with a detailed itinerary in “an extremely professional way” within “15 to 20 seconds.”
One U.S. interviewee, while sharing how they discuss AI chatbots with their college-age child, summed up the general sentiment of many interviewees, who acknowledged that AI chatbot use could involve tradeoffs between efficiency and accuracy:
It’s a great tool, but it’s not 100% gospel. It’s just taking the information that’s available on the web. So use it to help you kind of do those tasks that take time. but don’t have it write your paper for you or don’t have it do all your work for you. Use it as a tool but do your own work basically.
🇺🇲🇮🇳 Interviewees use AI chatbots to tailor content to their desired level of understanding.
Interviewees in both countries actively modulate the complexity of information to suit their expertise. Several interviewees ask their preferred AI chatbots to simplify further or provide additional detail, allowing them to “zoom in or out” to exactly the right level of detail for their needs.
An AI chatbot’s ability to generate information at varying levels of expertise is really attractive to the interviewees. Several Indian interviewees who explicitly mentioned this feature were mostly looking for information to teach their children or students, or they just wanted something explained to them in simpler language. U.S. interviewees also take advantage of this feature, especially to help them study. One of them asked ChatGPT to tell them everything that is likely to be tested on a particular physics concept, and then they noted,
When I am even having challenges to process all this information at the same time I just say: ‘I have an exam in 10 minutes. Can you just simply explain it to me easily,’ because I just try to simplify everything in a very short period of time.
🇺🇲🇮🇳 Interviewees like interacting with AI chatbots because they provide encouragement and an upbeat tone.
Many interviewees like the consistent affirmation of AI chatbots, particularly in situations where there might be a social risk associated with asking other people the same question. Specifically, several interviewees in both countries appreciate that the AI chatbot never judges them for questions they feel others might think are boring, stupid, or “a burden.” For one U.S. interviewee, the AI chatbot serves as a confidential problem solver. They recounted a time when using Copilot to navigate a website with unclear menu options, and noted: “it was nice having a Copilot with me to feel this out, okay? Because I don’t think I’m going to admit to a human that I couldn’t figure it out.” Similarly, an interviewee from India described the AI chatbot as a “brother” or “guru” specifically because of its infinite patience with “irritating” questions that they could not ask their friends and family repeatedly. In both countries, the AI chatbot’s consistent, non-judgmental and upbeat tone creates a space where interviewees feel comfortable asking “stupid” questions or sharing personal struggles.
While an AI chatbot technically does not hold conversations, but rather outputs statistically likely responses based on inputs and patterns from massive training datasets, nearly all interviewees used anthropomorphic language when describing their relationship with the AI chatbot and/or while using it.
Many interviewees said using an AI chatbot can feel like talking to a person and described them as “friendly” or having a “friendly demeanor” in their interactions. Others called the AI chatbots a “companion,” a “friend” or a “confidant.” We also heard interviewees describe AI chatbots as both supportive and educational, referring to them as a “coach,” a “personal assistant,” a “mentor,” a “teacher” and a “personal tutor.” AI chatbots’ use of positive reinforcement is particularly impactful. Multiple interviewees appreciate when the chatbot affirms the quality of their inputs or offers compliments, such as responding to their prompts with “great question.”
Even the interviewee who expressed a highly sophisticated understanding of how AI chatbots work still tended to use anthropomorphic language. The same interviewee acknowledged, “I know I don’t have to be polite to it. They don’t care. AI don’t got feelings” but still say “please” and “thank you” in their interactions because “it’s like a friend but like your personal assistant and mentor at the same time.” They were not alone in this style of interaction: at least a few interviewees in each country regularly say “please” and “thank you” during their interactions with AI chatbots.
Complicating the story
A very small number of interviewees actively resisted anthropomorphizing AI. The starkest example was one U.S. interviewee, who said, “For things like this, ‘help me figure this out’ [or] ‘help me search the internet’? Excellent companion. But as a human friend? It’s not.”
🇺🇲🇮🇳 Interviewees expressed concern about privacy and surveillance, but their behavior does not strongly reflect it.
Interviewees in both countries expressed concerns that their prompts or inputs could be shared or leaked. Their strategy for managing privacy concerns was generally to avoid inputting medical or financial information or identifying personal details like date of birth. Two interviewees in the U.S. worried that the full history of their queries could be identifying or compromising even without specific information like their full name. They did not know how their data was being used, but they did not take additional precautions beyond omitting specific details from their inputs. In contrast, other interviewees freely provided identifying information to AI chatbots, including birth date, full name, specific location and detailed medical documents.
Share
