Rich countries and fast-growing ones seem to be falling in and out of love with AI on different timelines — why?

This explores why trust in AI might rise or fall differently in wealthy countries than in fast-developing ones. Be aware up front that the collection doesn't contain a study that tracks this split directly. The closest source is a 47-country survey by KPMG and the University of Melbourne, which finds that trust is the strongest predictor of whether people accept AI. It also finds that trust is falling worldwide: perceived trustworthiness dropped from 63% in 2022 to 56% in 2024 Why does AI trust keep falling even though it matters most?. The collection doesn't break that decline down by country. What it does offer are several mechanisms that could produce different trajectories in different places, and these are more useful than a single headline number.

The first mechanism is the stage of adoption. Anthropic's Economic Index finds that AI use follows income. Wealthier countries use AI for a broad mix of tasks, while lower-income countries concentrate on coding. As adoption matures, use shifts from handing off whole tasks toward working alongside the AI Does AI adoption follow wealth and mature over time?. That matters for trust because long exposure is what erodes it. Stack Overflow's developer survey shows use climbing to 80% while trust in accuracy fell from 40% to 29%. The main cause was code that looks right but contains subtle errors, which leaves developers checking everything Why do developers keep using AI tools they don't trust?. One plausible reading is that people in advanced economies are further along that curve and have already met the disappointments that people in newer markets haven't reached yet.

The second mechanism is cultural. In one writing study, Indian writers accepted more AI suggestions than American writers. The authors argue this reflects different norms around trust and adopting technology, not a quirk to be filtered out of the data Is higher AI use by Indian writers a confound to control?. A cross-language study adds an important limit: in every language tested, people trust AI that sounds confident, even when it's wrong Do users worldwide trust confident AI outputs even when wrong?. So culture may change how much people rely on AI, but the underlying weakness is the same everywhere. Trust follows how confident the AI sounds, not whether it's right.

This points to a less comfortable explanation. Much of what we call trust isn't a judgment about accuracy at all. ChatGPT users trust it because it responds quickly and conversationally, which triggers the same social instincts people use with other people Does conversational style actually make AI more trustworthy?. Trust tends to recalibrate only when people repeatedly see the actual results Does revealing AI identity help or hurt user trust?. It drops sharply when an AI's mistakes become irreversible and public, such as an email that has already been sent, rather than merely high-stakes What makes people distrust AI agents they delegate to?. News behaves the same way: across 45 markets, people who use chatbots trust them far more than people who don't Does trust in AI chatbots drive news-seeking behavior?.

Taken together, the different trajectories may say less about whether AI is trustworthy and more about where each population sits on a shared cycle. Early on, conversational fluency and novelty build trust. Later, visible mistakes accumulate and trust erodes, even as use keeps growing. If that reading is right, high trust in emerging markets isn't necessarily a sign of better-informed users. It may be the same pattern at an earlier point on the curve. To test that, you'd need country-level data over time, which this collection doesn't have yet.