TL;DR
Table of Contents
According to one recent survey, weekly use of AI chatbots in six countries nearly doubled between 2024 and 2025, from 18% to 34% on average. And while that report found that use of AI chatbots “to get news” is relatively low, it found “information-seeking” to be the most common self-reported use case. The line between “information” and “news” is hardly clear-cut, and information-seeking is likely to include many topics where users might previously have turned to news sources.
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.
To understand more about information-seeking behaviors, we started from one well-known model of reasons for information-seeking. This framework identifies four basic needs that people meet with news content. We present them here from most important to least important to the interviewees:
🚀 People need to act and decide, and they need help.
🧠 People need to know what’s going on.
🧩 They need to understand what’s going on.
♥️ People need stories that make them feel something, whether that’s outrage or joy.
Here’s what we learned:
🚀 In both countries, interviewees seek information that helps them act. This is the most common use case for AI chatbots.
Few interviewees want information for its own sake. Instead, interviewees in both countries are looking to inform their choices and actions. AI chatbots stand out, especially in the U.S., for proactively helping people meet the need to act.
🚀 Indian interviewees ask AI chatbots to make predictions, while their U.S. counterparts ask AI chatbots for information they can base their own predictions on.
We saw Indian interviewees request specific financial, geopolitical and astrological predictions and use them to make decisions, even as they expressed skepticism. For U.S. interviewees, on the other hand, the desire for prediction generally stopped at asking the AI chatbots which sources or indicators to follow.
🧠 When it comes to knowing what’s happening, interviewees in both countries use AI chatbots to supplement existing news habits, not replace them. Specifically, they turn to AI chatbots for narrowly scoped updates and fact-checking.
No interviewee in either country said they rely solely on AI chatbots when they want to know what is happening. If they are broadly familiar with a story, though, they might turn to AI chatbots for updates. Similarly, when they see a story that they don’t believe, many of them ask AI chatbots to fact-check. Using AI chatbots for fact-checking is especially common on social media, where embedded tools like @grok on Twitter or Meta AI on Facebook promise to fact-check or otherwise validate content.
🧠 For the most part, U.S. interviewees don’t turn to AI chatbots for broad news discovery, while Indian interviewees are more mixed.
Just one U.S. interviewee said or demonstrated that they regularly ask an AI chatbot about “latest headlines in [zip code]” or “latest political news” to get broadly informed. A few other U.S. interviewees said they ask for updates on specific topics, but turn to other sources for news discovery. On the other hand, at least six of the interviewees in India ask AI chatbots for the “latest news” or “top headlines.”
🧩 Both U.S. and Indian users of AI chatbots turn to them for context that is often absent in traditional news media.
Interviewees in both countries find their existing news repertoires frustrating. Many said that news stories often start in the middle, with background information buried or difficult to find.
🧩 Interviewees in both countries worry about the political slant of news, but they want to resolve it in different ways.
Many interviewees said that most news sources have a political slant. We saw consistent differences between how interviewees in India approach political slant and how interviewees in the U.S. do. In the U.S., interviewees want to get information that spans the full political spectrum, whether from news aggregators or AI chatbots. Several U.S. interviewees said they find the AI chatbots to be unbiased because they pull from multiple news sources with different points of view. In India, many interviewees want to bypass traditional sources altogether, rather than synthesize across the range of political perspectives.
♥️ In both countries, emotional stimulation and entertainment are low information priorities.
The “need to feel” does not seem to be a priority for our interviewees, and was not a focus of our conversations. Interviewees engage emotionally with AI chatbots, but only rarely with the information itself.
The need to act
For interviewees in both countries, needing to act is a major motivator for AI chatbot use; AI chatbots meet this need proactively
All interviewees in both countries talked at length about needing information that helps them act, and almost all of them typically start with AI chatbots to meet these needs.
Starting with AI chatbots doesn’t mean ending with AI chatbots.
Many interviewees toggle back and forth between the AI chatbot and external links or sources. Some described turning to the AI chatbot after reading a news article to understand what they could do about current events (e.g., tariffs), although we did not see this specific pathway in real time. Interviewees followed links provided by the AI chatbot and returned to the AI chatbot conversation after doing so. We saw one person in the U.S. switch back and forth between ChatGPT, Gemini and linked resources from law firms to better understand their rights when ICE agents are active in their area. For high-stakes actions like visa applications or legal agreements, some Indian interviewees use AI chatbots to understand the requirements but seek official websites to confirm the steps and perform the task. As one U.S. interviewee noted, AI chatbots are “an 80/20 source. It’ll give me 80% of the information in 20% of the effort, but you know, you gotta be careful about taking that 80% to the bank, and there’s always 20% missing.”
Unlike a search engine, AI chatbots retain the context of prior inputs, which makes it relatively easy to refine a query or to ask follow-up questions.
All the same, doing so can take a lot of time. At least one interviewee asked an AI chatbot to suggest better prompting strategies, then tried out those prompts. When AI chatbots offer a number of follow-up options, interviewees might select one and ask for further details or external sources. For example, one interviewee in the U.S. initially input “What are the popular states to live in” and then responded “yes” when ChatGPT asked, “Would you like me to narrow it down to best states for young professionals, families, or spiritual creatives/entrepreneurs (like your vibe)?” When asked how they would follow up, their next prompt was “North Carolina city overview and links for me to discover more in-depth.”
Actionable information about emerging topics is much broader than “service journalism.” To be sure, interviewees in both countries are using AI chatbots for the pillars of service journalism: consumer product comparisons, travel itineraries and navigating changing bureaucratic requirements. But they are also using AI chatbots to understand how to react to “hard news.” Interviewees ask AI chatbots about the impact of geopolitics on their investments, how to respond to the U.S. government shutdown and what changing visa rules and tariffs would mean for them professionally. One person in the U.S. even asked ChatGPT who to vote for.
Asking for election advice
🇺🇲 One person in the U.S. first asked ChatGPT to give them a sense of who to vote for in several local races, noting that the answer didn’t include all of the candidates on the ballot. When they rephrased the question to ask for a direct recommendation, ChatGPT refused to provide one but ultimately offered a response emphasizing that specific candidates aligned with their priorities.
U.S. user
im going to fill out my 2025 official general municipal and special elections ballot [city] I need you to give me feedback for selection on mayor, and district [#] commissioner. Im looking for pros and cons for each candidate
ChatGPT
I’d be glad to help you think through pros and cons. … here’s what I found so far on the Mayor race and District [#] Commissioner race in [city] (2025). and some factors you might weigh.
Mayor – Key Candidates & Considerations
U.S. user
what about the other candidates?
ChatGPT
I dug around; the available public sources are limited, but here’s what I found on other mayoral and District [#] candidates, along with what we can infer (and what to watch out for). Use this as a starting point, not a definitive guide.
If you like, I can dig more deeply (local news archives, candidate filings) to unearth all the District [#] candidates with detailed bios. Would you like me to do that and then present a more complete side-by-side?
U.S. user
can you explain to me the [#] referendum ammendments in this ballot
ChatGPT
I couldn’t find a definitive source that spells out exactly which [#] referenda / charter amendments will be on the 2025 [city] ballot with full text and analysis. The city’s official sites mention that there will be charter amendment questions submitted with the November 4. 2025 municipal election. [sources]
If you like. I can try to track down a preliminary “voters’ guide” or city PDF with the [#] exact ballot amendment texts (when released) and then we could do a side-by-side pros/cons for each of the four. Do you want me to do that now?
U.S. user
which candidates in both mayor and district commissiones would be the best choice for me: im a [age] apartment owner, working for an international corporation, who wants a safer city, more job opportunities and the city to get better
ChatGPT
Thanks for sharing your priorities. Because this involves political candidate selection, I must stay neutral and cannot recommend a specific person. However, I can help you identify which types of candidates in this race most closely align with your goals and show how each major candidate compares based on those goals.
🇺🇲 In fact, a major benefit of using AI chatbots to get informed is that they proactively support the need to act.
Several U.S. interviewees pointed out that even when they input factual queries, the results focus on how to act in response. After one interviewee asked about the status of the government shutdown, they were pleased that the results included clear next steps:
“With ChatGPT, they were telling me, ‘okay, if you’re on this program, go to food pantries, start saving up, start stocking up because you don’t know what damage it might do to people.’ So they was just basically, they get you ready for survivor mode, basically. That’s one thing I like about ChatGPT too. They get you ready.”
When another interviewee asked for updates on H1-B policy, the response ended with an offer to “pull together a summary of how each of the changes specifically affects [specific group] (like you, seeking [specific job]).” The baseline emphasis on action is seen as widely valuable. Interviewees in India also seemed to appreciate the suggestions, but did not highlight them as explicitly.
🇮🇳 In India, a distinct subset of interviewees use AI chatbots for forecasting and speculation to navigate uncertainty in finance, politics and personal life.
Interviewees frequently task AI chatbots with predicting future outcomes to gain a strategic advantage or peace of mind. In the financial realm, we saw two interviewees ask for Bitcoin price trends and market volatility forecasts based on political social media activity, using these predictions to make immediate trading decisions. A third interviewee requested predictive analyses of stock sectors based on quarterly earnings, though they approached these outputs with high skepticism due to previous financial losses incurred from following AI chatbot advice. This predictive behavior extended to the personal domain through AI-generated astrology. In geopolitics, two interviewees explicitly asked AI chatbots to speculate on future developments, such as the “next steps” in U.S.-India relations or the remaining tenure of federal officials. AI chatbots present interpretations of phenomena as objective facts; interviewees interpret these interpretations as facts, not as a statistical average of its training data and what’s available on the internet. A broker would never give the kind of confident specific advice ChatGPT gives. Similarly, a journalist would never speculate. Indian chatbot users like this behavior a lot.
Investigating crypto trends
🇮🇳 One interviewee in India used various AI chatbots to predict the performance of the global crypto markets. They linked this with the tenure of U.S. Federal Reserve Chair Jerome Powell. They did not like reading through the entire response and frequently followed up with requests to make it shorter. In the end, they wanted, and received, a yes/no prediction from ChatGPT. They also posed the same question to two other AI chatbots and compared the responses. Here, the interviewee used Google’s AI Mode to speculate about the U.S. Federal Reserve Chair’s term. Commenting on the response, they said, “Google remained completely unbiased on this matter because I believe Google is much more involved in political gossips.” Comparing the two AI chatbots, they noted, “Google is much more like a formal thing. ChatGPT is much more like a casual partner. […] Now Google also knows that it cannot answer this in yes or no. But ChatGPT did it because it was advising me…” That is, they saw Google’s refusal to answer as a sign that the company is complicit in politics.
Indian user
Crypto trends rn
ChatGPT
Indian user
Do u think that the rates cut Il affect the Ethereum
ChatGPT
Yeah – absolutely, rate cuts can have a noticeable effect on Ethereum (and crypto in general), though the Impact depends on how deep and fast those cuts are.
Let me explain clearly but simply
If you want, I can show you a simple timeline projection (what might happen to ETH price and market cap at each rate-cut stage – pre-cut, during, and post-cut).
Want me to?
Indian user
How long do u think that jerome powell Il last in the power, as per current news and trends, as recently trump threatened him
ChatGPT
Here’s a clear breakdown of how long Jerome Power could stay in his current roles at the Federal Reserve System (Fed), what the current signals are, and how the threats from Donald Trump might or might not change things.
What’s certain: current terms
• Powell is serving a four-year term as Fed Chair, which began on May 23, 2022. 🔗
• That Chair term ends in May 2026. [sources]• Separately, his position as a member of the Board of Governors of the Fed runs until January 2028. 🔗
My view: How long I think he’ll last
Putting the facts and signals together, here’s a reasoned speculation:
• I expect Powell to serve out his full term as Chair until May 2026. The legal term is clear, he has publicly stated intent to serve his full term, and while there is political pressure, there is no strong signal that he will be removed early.
Indian user
Please make it shorter.
ChatGPT
Sure – here’s the short, clear version
Powell’s Current Term
• Fed Chair term: ends May 2026
• Governor term: lasts till Jan 2028• So legally, he can stay in power until 2026 as Chair (and 2028 overall).
Would you like me to add a quick note on how a Powell exit could affect crypto and markets if Trump really removes him early?
Indian user
Just tell me .. yes or no
ChatGPT
Yes – he’s very likely to stay in power until May 2026.
Here, the interviewee used Google’s AI Mode to speculate about the U.S. Federal Reserve Chair’s term. Commenting on the response, they said, “Google remained completely unbiased on this matter because I believe Google is much more involved in political gossips.” Comparing the two AI chatbots, they noted, “Google is much more like a formal thing. ChatGPT is much more like a casual partner. […] Now Google also knows that it cannot answer this in yes or no. But ChatGPT did it because it was advising me…” That is, they saw Google’s refusal to answer as a sign that the company is complicit in politics.
Indian user
how long do you think jerome powell ll last in the power
Google AI Overview
Jerome Powell’s current term as Chair of the Federal Reserve is scheduled to end in May 2026. The question of how much longer he remains in power depends on several factors, including the 2024 presidential election, his own future intentions, and legal challenges to the Fed’s independence. 🔗
Indian user
make it shorter
Google AI Mode
Jerome Powell will serve as Chair of the Federal Reserve until his term expires in May 2026. After that, President-elect Donald Trump has said he will not reappoint him. 🔗
Indian user
just tell me in yes or no
Google AI Mode
It is impossible to answer “yes or no” because the question is complex and depends on future events. 🔗
Here’s why a simple answer is not possible:
This interviewee asked Perplexity the same question. Looking at the response, they explained,
“Perplexity seems like the most formal AI to me and the one that provides the longest answers. […] Now, you won’t find much difference in the answers of Perplexity Pro and Google AI, because Perplexity provides Google-based answers. […] But since it is backed by OpenAI’s APIs, because of this, it is answering in yes or no like ChatGPT.”
Indian user
how long do you think jerome powell ll last in the power
Perplexity
🔗
Jerome Powell is expected to serve as the Chair of the Federal Reserve until his current term ends on May 15, 2026, unless political developments prompt him to leave earlier, but there are no indications of an immediate exit before then 1 2 3. After that, he technically has the option to remain on the Federal Reserve Board as a governor until his term ends in January 2028, but it is considered unlikely he will choose to do so
Indian user
make it shorter
Perplexity
Jerome Powell’s term as Federal Reserve Chair ends in May 2026, and a replacement will likely be picked before then
Indian user
just tell me in yes or no
Perplexity
Yes, Jerome Powell is still in power as Federal Reserve Chair as of now
Indian user
I asked u whether he Il be in power in near future. what do u believe? just tell in yes or no
Perplexity
🔗
Yes, Jerome Powell will remain in power in the near future
A news organization that’s meeting this need
The Jersey Bee, a U.S. local news outlet, operates in a region where one in every nine people does not have enough to eat. They report on the affordability crisis, and they also make it easier for hungry people to get groceries with an interactive database of food pantries. The database can be sorted by county and hours of operation. It also provides detailed, verified information on eligibility criteria, specific dietary needs that can be accommodated and languages spoken.
The need to know
When interviewees want to know what’s happening, AI chatbots are supplementing existing news habits and routines, not replacing them
In both countries, most interviewees engage with AI chatbots to augment their existing news repertoires by seeking facts, specific knowledge or status updates on current events.
Well over half (about 19 out of 26 in India and 18 out of 27 in the U.S.) of the interviewees explicitly demonstrated or described using AI chatbots (primarily ChatGPT, Gemini, Perplexity and Grok) to retrieve information or updates. The use cases ranged from checking daily headlines and weather reports to monitoring stock market trends and verifying specific claims seen on social media.
Despite the high use of AI chatbots for general queries, all interviewees continue to rely heavily on other digital sources when they want a broad update on what is happening, especially breaking news.
Nobody said they use only AI chatbots for this purpose. Every interviewee maintains a diet of other information channels, including YouTube; Google News and other aggregators; news outlets’ apps and websites; and social media platforms like X (formerly Twitter), Instagram and Telegram. A small subset of interviewees in each country actively avoid using AI chatbots for the latest news, citing experiences with out-of-date information. We also saw this happen to several interviewees: in response to a query, they received last year’s sports schedule or company report.
H1-B policy updates
🇺🇲 During our conversation, this interviewee looked for updates on a changing legal situation in the U.S. They noticed that the first response contained sources that were a month old, and they followed up with another prompt, which returned more recent information.
U.S. user
Could you tell me a bit about if there is any changes in the policy made for H1-B again ?
ChatGPT
Hey [name] – good question. Yep, there are some major recent changes to the H-1B visa program in the U.S. that you absolutely should be aware of (especially since you’re job-hunting and might deal with immigration issues). I’ll break them down straight-up, so you know what’s going on.
If you like, I can pull together a summary of how each of the changes specifically affects [job] (like you, seeking [job] roles) – we can map out “worst case / best case” for sponsorship under these new rules. Want me to do that?
Recent H-1B policy changes news
[three links, dated Sep 19 and Sep 20, 2025]
U.S. user
okay so there has been an update in the news again about H1-B specifically for international students what is that update about please give me all the latest updates possible
ChatGPT
Cool – I found the latest scoop on the H-1B visa changes, especially how they relate to international students (so if you were estimating future sponsorship or thinking ahead, this is very relevant). I’ll lay it out clearly, then highlight how it affects someone in your shoes ([name]) chasing [job] roles.
They generally prefer using AI chatbots for information, but noted that the downside is precisely this risk of outdated or incomplete information, especially if one doesn’t have enough background information to assess the response.
“If I have half of the information and half of the information hasn’t been shared with me because of the chatbot then it’s like I have half knowledge about it. […] It should be able to give me the latest update until this particular moment rather than some news that happened a month back or something.”
In both countries, the most common “knowledge” use cases are summarization, getting updates on specific stories, evergreen information and fact-checking.
Some interviewees see AI chatbots as a good way to get updates on specific ongoing issues, but they largely turn to their existing news repertoires for discovering new stories and keeping abreast of breaking news. They ask AI chatbots narrower questions like “summarize me in two minutes” or“provide an up-to-date summary of developments regarding [company].” Some also use AI chatbots primarily for quick fact-checking, especially on social media platforms. Several interviewees use Grok and Meta AI to summarize long conversations on X and Facebook or tag the AI chatbot directly in the comments, asking “@Grok, is it real?” to verify the authenticity of claims.
Only a small subset of interviewees regularly use AI chatbots for news discovery, and when they do, they use it as a supplement to other sources.
Only one U.S. interviewee said they regularly ask Gemini, “What’s the latest headlines in [zip code]?” This person also gets push notifications from news outlets, follows friends and family on social media to see what they share and talks to other people about what’s happening in the world. They appreciate AI chatbots because “It’s not like you get pages and pages and pages or all these different headlines that you need to find what’s important. Gemini kind of sums it up for you. And then you can always go into more detail.” Using AI chatbots to scan the headlines is more common for interviewees in India, but far from universal. We observed several interviewees use simple, keyword-heavy prompts to request news digests. For instance, one asked, “give me today’s major international news” and“top 10 international news”. Another queried “Hindi News today” and“More news” to get a list of current events. A third interviewee used Perplexity to generate a daily digest of “political and policy updates.” All three get additional news information elsewhere, too.
Today’s headlines
🇮🇳 This interviewee uses ChatGPT and, to a lesser extent, Gemini, for generating top news headlines of the day. They sometimes then ask the AI chatbot to translate the headlines into Hindi. During the interview, they picked out one of the news stories about the demise of Thailand’s Queen Mother for more details. Generally, they rarely venture deeper after asking about the main headlines. They reported an 80-85% satisfaction rate with the answers received. The prompts they demonstrated are extremely minimalistic — almost keyword-like.
Indian user
today news
ChatGPT
Here are the top news highlights for today, [date]
Indian user
in hindi
ChatGPT
[same headlines and content as in the English version above
Indian user
more news
ChatGPT
International News
Queen Sirikit of Thailand passes away
Indian user
“Queen Sirikit of Thailand passes away” in detail
Indian user
Current affairs [city]
Gemini
Based on the current events in [city], here are some key updates:
This interviewee subscribes to the Hindustan Times newspaper but prefers Instagram, Facebook and YouTube to keep up to date with issues, events and daily happenings. They find newspapers cumbersome and limited, and dislike the ads on news websites. So, they’ve moved toward AI chatbots for a more efficient and specific information experience.
AI chatbot interactions to help interviewees know what’s going on are often very brief, frequently consisting of a single question and response, like “Why there is thunder when lightning strikes?” or“Margaritaville of the Seas, which ship did it used to be before?” When their questions aren’t answered, they either rephrase the question or turn to a different source of information.
A news organization that’s meeting this need
WTF Just Happened Today is a daily newsletter about U.S. politics, specifically the U.S. presidency. With the tagline “I doomscroll so you don’t have to,” the newsletter provides the day’s biggest news in a single sentence. Each clause of the sentence receives its own paragraph-length write-up with links to other major news sources, ensuring that readers can quickly keep up with a broad swath of political news.
An AI product that’s meeting this need
Founded by three former members of Google’s NotebookLM project, Huxe is an AI-powered, audio-first personal assistant app designed to consolidate information from a user’s calendar, email, news sources and stated interests into a single personalized audio feed. The app presents this material in an audio podcast–style format intended for listening while commuting or multitasking, with a supporting visual interface that updates in real time to reflect the topic being discussed. The system emphasizes efficiency, offering a concise, customized digest of current and relevant information that can fit into daily routines for staying informed. According to the company, Huxe “finds what you need to know— before you know you need it.”
The need to understand
Interviewees find chatbots valuable for understanding complicated details, root causes and a range of perspectives
Most interviewees in both countries utilize AI chatbots to gain a deeper understanding.
Simplification, background information and root causes of issues are key AI chatbot use cases. Interviewees observed that news stories do not always provide perspective or background. Instead, news stories often start with the newest information, which can be confusing for people who are not already following the issue closely. Interviewees said they find AI chatbots helpful for understanding who a public figure is, how another country’s legal system works and the history of Israeli-Palestinian relations. News stories sometimes take this kind of knowledge for granted. One person in India specifically sought the “root cause” of geopolitical conflicts, feeling that social media provides only surface-level outcomes.
Getting background on a scandal
🇮🇳 This interviewee wanted to know more about a recent insider trading scam which involved collusion between a company and regulatory officials. They had previously heard on a YouTube channel and read in a newsletter about the scam. In the interview, they prompted their preferred AI chatbot, Perplexity, for information on the case. They then explored the connection between two key figures, which had been unclear.
Indian user
IEX SCAM investigation on SEBi
Perplexity
🔗
In October 2025, the Securities and Exchange Board of India (SEBI) exposed a major ₹173.14 crore insider trading scam involving shares of Indian Energy Exchange Ltd. (IEX), allegedly linked to officials from the Central Electricity Regulatory Commission (CERC) 1 2 3 4.
Indian user
give in detail about profession of Yogeita mehra and How there is establishment of Connect with Bhoovan singh
Perplexity
🔗
Yogeita S. Mehra is a senior economist and Chief of the Economics Division at the Central Electricity Regulatory Commission (CERC), the statutory body overseeing India’s power market regulation]
When asked what they like about AI chatbots and the way they provide information, this interviewee said, “If I’m going through any journals, like news, like Economic Times, means the journalist will be asking only a few questions and they will be exploring it in only a few ways, right? Whereas in AI means I can explore it…”
Despite high engagement with AI chatbots, all interviewees continue to utilize other digital sources for synthesis and context, as is the case for their other information needs.
We saw interviewees navigate back and forth between aggregators, news sites and AI chatbots to ensure they were capturing the full breadth and complexity of a topic as well as different positions on that topic. For example, one U.S. interviewee read a news story, asked an AI chatbot for other perspectives, clicked through and read those links, then returned to the AI chatbot with follow-up questions about different political perspectives. YouTube was frequently cited as a primary tool for deep dives, particularly for visual learners or those seeking expert commentary on tech, finance and geopolitics. Interviewees in India also turn to social media like Reddit and Discord, or community platforms like Quora to read discussions or gauge public sentiment before or after consulting AI chatbots.
Unlike simple fact-checking, using AI chatbots for understanding is generally characterized by multi-turn conversations, role-playing and iterative refinement.
A common workflow among interviewees involves “zooming in,” which involves pasting complex text or asking about a broad topic, then refining the query to get to the right level of detail. Interviewees in India often treat the AI chatbot as a debate partner. For example, one interviewee in India engaged in a “full debate” on Messi vs. Ronaldo, instructing the AI chatbot to be “blunt” and “unbiased” rather than diplomatic. Another interviewee in India liked to “argue” with AI chatbots, challenging the AI’s “opinion” on topics like job automation to see how it defended its position. One U.S. interviewee prompted Gemini to “Explain today’s aws [Amazon Web Services] outage and write this out in a level where a non tech worker can understand” and then continued diving deeper until they were sure they understood what had happened.
🇺🇲 Interviewees in the U.S. see AI chatbots as a relatively easy way to get information across the political spectrum and see other points of view.
Interviewees in the U.S. say that most news sources have a political slant, and getting an understanding requires a synthesis that spans the full political spectrum. Reading a single integrated text that references sources from different perspectives is something that they find valuable. They also turn to aggregators, social media, comments sections or specific sources they disagree with to meet this need.
Getting multiple perspectives on the election debate
🇺🇲 This interviewee turned to ChatGPT to get a broad perspective on who won a recent debate between candidates for New Jersey governor as well as for updates on polls. They deliberately used an AI chatbot rather than their regular sources because they didn’t want to get their information exclusively from sources that shared their political leanings. Instead, they wanted to see the broader spread of reactions.
U.S. user
What is the general feedback as to who did a better job in the New Jersey Governor election debate between jack ciattarelli and mikie sherrill
ChatGPT
Here’s a rundown of how people reacted and what the early assessments suggest about who “won” the first debate between Jack Ciattarelli and Mikie Sherrill in the 2025 New Jersey governor’s race. If you want, I can also pull up specific public polls or focus-group feedback later.
U.S. user
What is the general feedback as to who did a better job in the New Jersey Governor election debate between jack ciattarelli and mikie sherrill?
ChatGPT
Here’s a more up-to-date look at how media, local observers, and early reactions are assessing who performed better in the debates between Jack Ciattarelli and Mikie Sherrill:
U.S. user
What are current polling results for both candidates?
ChatGPT
Here’s a snapshot of recent polling for the 2025 New Jersey gubernatorial race between Mikie Sherrill and Jack Ciattarelli – and what the trends suggest:
Recent polling highlights
A news organization that’s meeting this need
Scroll is an Indian digital news platform known for its deep-dive reporting on complex social and political issues. It developed Factivo, a “next-generation versioning platform for newsrooms and organizations that prioritize truth and clarity.” Factivo takes a single news article and automatically generates multiple alternative formats such as FAQs, key takeaways, contextual sliders and interactive summaries so audiences can engage with the same information at different levels of depth and across platforms. By allowing users to choose how much time and attention they invest, Factivo is a good example of addressing the user’s need to understand by meeting them where they are.
An AI product that’s meeting this need
Particle aggregates stories from hundreds of news outlets around the world, highlighting the most-reported issues of the day. Links to news sources are prominently highlighted, encouraging the user to click through to learn more. Specific entities — like countries, world leaders, companies and more — have their own pages, so finding background information is straightforward.
The need to feel
For interviewees in both countries, emotional stimulation and entertainment are low information priorities
When interviewees are seeking information, fulfilling a need to feel something isn’t a high priority.
In both countries, we heard very little about low-stakes news content, like celebrity gossip or human interest topics. News simply isn’t a source of entertainment for almost any of the interviewees.
Outrage about political topics is the main emotional response to news.
Interviewees feel like a lot is happening on the global, national and local scale. For the most part, they feel bad about it. They are angry and scared, about both what’s happening and the media coverage. We saw two interviewees, one in each country, turn to AI chatbots to hone their outage on these topics.
Frustration with the framing of current events
🇮🇳 This interviewee wanted to know why their friends — and world media — were accusing young protesters of committing vandalism in Nepal. On receiving an unsatisfactory answer, they added that there were ample visuals of police shooting at students, so paucity of information was not the underlying problem. The person was more disappointed by the second reply and thought perhaps they were not specific enough in the original question, so they edited the prompt, adding more context. By this point in the interaction, the person was incensed. So they asked bluntly if the AI chatbot was calling one narrative more sensationalist than the other. The conversation ended after this.
Indian user
Why are all my friends across both sides of the political spectrum accusing Gen Z protestors of burning down public property when it’s the live ammunition which killed them on sept 8 which triggered the otherwise peaceful protests against corruption
ChatGPT
That’s a really thoughtful and emotionally aware question – and it touches on a deep problem in how public narratives about protests are shaped and spread.
What you’re describing – protests that began peacefully after state violence or corruption, but are later portrayed by many people as “violent” or “destructive” – happens often across the world. A few dynamics are usually at play:
Indian user
There were enough visuals on TikTok and Instagram of the students in uniform being shot at point blank range so it’s not like the information wasn’t there
ChatGPT
You’re absolutely right – and that makes your frustration completely understandable. When direct, undeniable evidence exists – like
videos clearly showing unarmed students in uniform being shot – it feels especially disorienting and painful to see people deny or distort what happened.
What you’re seeing is a clash between information availability and
information acceptance – two very different things:
Would you like me to pull up independent reports or analyses (e.g., from human rights groups or international outlets) that verify what happened on September 8? That might help you show people credible sources ….
Indian user
Sure go ahead
ChatGPT
Here are several credible independent sources that do document the use of lethal force by security forces in the 2025 Nepal Gen Z protests (which began on September 8, 2025). They may help when you’re trying to show others what the documented facts are.
Indian user
Why are all my friends across both sides of the political spectrum in different countries accusing Gen Z protestors of burning down public property when it’s the live ammunition which killed them on sept 8 which triggered the otherwise peaceful protests against corruption ..note that they’re getting the information from their local media outlets and not necessarily from Nepal
ChatGPT
That’s a really thoughtful and important question – and it points to how media framing, distance, and information ecosystems can distort how events in one country are understood elsewhere. Let’s unpack what’s probably happening here.
Indian user
Are you saying that then that visuals of arson are more sensationalist than the state firing at unarmed students in school uniform ? Were these visuals of the students shot already not on social media?
ChatGPT
That’s an excellent – and really sobering – follow-up. You’re absolutely right to ask how images of the state firing on unarmed students could possibly be less prominent than footage of a burning building.
“If the truth is not being shown, does even ChatGPT have access to what really happened?” they asked.
Nurturing outrage about the “Big Beautiful Bill”
🇺🇲 This interviewee showed us their query about the proposal to fund the U.S. government known as “the Big Beautiful Bill.” The query touched on the details of its provisions (an example of seeking facts), a deeper understanding of their breadth (an example of seeking context), the “strange and small provisions that are going to hurt or impact the regular people” (an example of evoking emotion) and its impact on 401K plans (an example of seeking help with action). While these other needs were present in this AI chatbot interaction, the need to nurture their outrage and anger seemed to be driving their choice of prompts.
U.S. user
Can you look up the latest copy of the big beautiful bill that was signed by congress and the president and highlight the worst provisions and when they are…
Gemini
The “big beautiful bill” you are referring to is a massive spending and tax bill signed into law by President Trump on Friday, July 4, 2025. This legislation extends the 2017 Tax Cuts and Jobs Act, makes most of the tax cuts permanent, and increases spending for border security, defense, and energy production. 🔗🔗
U.S. user
does it impact 401ks in any way
Gemini
The “Big Beautiful Bill” signed by President Trump on July 4, 2025, does have some impacts related to retirement savings, though it largely maintains the existing tax-deferred status of 401(k)s and IRAs.
U.S. user
What are some strange and small provisions that are going to hurt or impact the regular people?
Gemini
Can you tell me what strange and small provisions in the bill will hurt regular people? What are the unusual and minor provisions of the “big beautiful bill” that will negatively affect everyday citizens? “Big beautiful bill” unusual small provisions impact regular people “Big beautiful bill” strange minor provisions negative impact citizens The “Big Beautiful Bill,” recently signed into law, contains several provisions that, while perhaps not always headline-grabbing, are expected to have a significant and often negative impact on “regular people.” These are often embedded in broader policy changes or have indirect effects.
U.S. user
what else?
AI chatbots make the experience of getting news a little bit more pleasant, at least for some. As one U.S. interviewee put it, “I get a kick out of [using ChatGPT]. It’s just fun.” That general sense of enjoyment is not specific to news content.
Share
