Product1 publisher3 min readPublished
Gallup finds AI optimism running ahead of trust in 35 of the 37 countries it surveyed
Gallup and Microsoft's first 37-country cut puts awareness, daily use and trust in accuracy in very different places. The spread lands on whoever has to ship one onboarding flow into all of those markets.
The Product Desk · Product desk
What happened
- Gallup published the first 37 countries of an AI engagement study run with Microsoft on Tuesday, and the eventual sample is set to reach 140 countries.
- Across those 37 countries, the median share of adults aware of AI was 81 percent, while the median share who had actually used it was 43 percent.
- Singapore led both measures, with 96 percent of adults aware of AI, 79 percent having used it and 46 percent using it daily; in Malawi 12 percent were aware and 1 percent used it every day.
- The United States ranked first for worry, with 74 percent of AI-aware adults saying the technology makes them feel worried.
Compiled by The Product DeskSomething wrong?How this is made
Why it matters
- constraint One worldwide onboarding flow is written for the median user: it over-explains to markets where daily use is already habitual and under-explains to the large group that knows the category but has never opened a model.
- decision A global product owner has to choose between per-market verification defaults and a single flow; the first costs a shared release train, the second drags on markets where trust is already high.
- exposure An American internal rollout depends on the same heavy users who are most likely to say their own jobs will be eliminated.
- contradiction Gallup's benefit scores and trust scores point different ways almost everywhere, so a market that screens as favourable on sentiment can still decline to act on what the model produces.
A global assistant ships with one first-run screen, translated. Gallup's medians say that screen is landing on two unrelated audiences in the same release. Subtract median use from median awareness and you get 38 points of adults who know what AI is and have not used it [1]. That group needs a first completed task. The daily users need somewhere to put a correction.
The trust column is where a rollout gets expensive. A median of 36 percent of AI-aware adults said they trust AI completely or a lot to provide accurate information, and only in Israel, Vietnam, China and Nigeria did that reach a majority [5][6]. Expected national benefit ran well ahead of trust in accuracy in 35 of the 37 countries [7]. China topped the benefit question at 93 percent, Vietnam at 91 percent and Singapore at 77 percent [19].
The American numbers separate cleanly. Worry among AI-aware US adults ran 38 points above the 36 percent who said AI would mostly help their country [2][8]. The US sits at the top of the worry ranking, ahead of the Netherlands at 67 percent, Canada at 64, the United Kingdom at 63, New Zealand at 61 and Ireland at 60 [10]. Gallup called the pattern the "Paradox of the Worried West", with seven wealthy Western countries among the eight reporting the most worry [11]. Positive emotions outweighed negative ones in 34 of the 37 countries, the exceptions being the US, Egypt and Palestine [12].
For an internal rollout, one earlier Gallup finding matters more than the country ranking: US workers who use AI daily or multiple times a week are more than twice as likely to say their job is very likely to be eliminated within five years [13]. The knowledge-worker exposure theory does not travel well either, since the IMF estimates 77 percent of Singapore's workers are highly exposed to AI [14] and Singapore sits near the top of the benefit table.
A separate POLITICO poll conducted by Public First and published Wednesday found roughly half of adults surveyed in the US, Canada, the UK, France, Spain and Germany said AI is already good enough and supported pausing further development to reduce risks [15]. Gallup's own instrument measured awareness, frequency of use, feelings, expected impact and trust [17]. Those are self-reported answers. Retention, task depth and onboarding copy sat outside what it asked.
A two-column read per market gets you further than one global trust posture. The two local numbers are the share who have used AI and the share who trust it for accurate information. The first run answers one column or the other. Where use is high and trust is low, the work is verification: visible sources and an edit step before anything is sent. Where use is low, the work is one finished task, and an accuracy disclaimer is noise a user cannot yet interpret. The cost of splitting is real, because two flows mean two release trains and two help centres to keep current. The published table covers about a quarter of the eventual sample [3].
What to watch
- Gallup's later cuts toward 140 countries will show whether high worry tracks Western markets or high incomes generally.
- Whether Gallup publishes per-country tables pairing use with trust in accuracy, the pairing a market-by-market rollout plan needs.
- Whether the Public First finding on pausing development holds up in polling outside those six countries.