R0031/2026-03-29/C002 — Assessment¶
BLUF¶
The claim conflates multiple sources. The "66% use AI" figure comes from the KPMG/University of Melbourne study (47 countries), not an Ipsos 31-country survey. Ipsos did publish a "use despite not trusting" finding, but from a US-only Consumer Tracker. The Ipsos 31-country survey (2023, 22,816 respondents) reports attitudes and nervousness but not the 66% usage figure. The underlying phenomenon is real but the attribution is incorrect.
Probability¶
Rating: Likely (55-80%)
Confidence in assessment: Medium
Confidence rationale: The phenomenon of high AI usage despite low trust is well-documented across multiple surveys. However, the specific attribution to "Ipsos, 31-country survey" cannot be confirmed. The 66% figure most likely comes from the KPMG/Melbourne study. Medium confidence because it is possible that a specific Ipsos report not found in this search does contain this exact statistic — but no evidence of such a report was found despite targeted searching.
Reasoning Chain¶
-
FACT: Ipsos conducted a 31-country survey in 2023 with 22,816 respondents. It found 54% excited, 52% nervous about AI. It did NOT report "66% use AI." [SRC02-E01, High reliability, High relevance]
-
FACT: Ipsos US published "People don't trust AI tools, but use them anyway" stating "two in three who use AI tools say they don't trust them." This is US-only data. [SRC01-E01, High reliability, Medium relevance]
-
FACT: The KPMG/Melbourne study (47 countries, 48,340 respondents, 2025) reports "66% of people use AI regularly" and "only 46% are willing to trust AI systems." [SRC03-E01, High reliability, High relevance]
-
JUDGMENT: The claim appears to conflate the KPMG 66% usage figure with the Ipsos brand and 31-country survey count. The "use despite not trusting" framing may come from the Ipsos US Consumer Tracker headline. The underlying phenomenon is real but the specific attribution chain is broken.
Evidence Base Summary¶
| Source | Description | Reliability | Relevance | Key Finding |
|---|---|---|---|---|
| SRC01 | Ipsos US Consumer Tracker | High | Medium | "Two in three don't trust but use anyway" — US-only |
| SRC02 | Ipsos 31-country AI Monitor 2023 | High | High | Does NOT report 66% usage figure |
| SRC03 | KPMG press release | High | High | Reports 66% usage + low trust (47 countries) |
Collection Synthesis¶
| Dimension | Assessment |
|---|---|
| Evidence quality | Robust — primary sources from both Ipsos and KPMG |
| Source agreement | High — all sources agree on the phenomenon, but disagree on where the 66% figure originated |
| Source independence | Independent — Ipsos and KPMG/Melbourne are separate organizations conducting separate surveys |
| Outliers | None |
Detail¶
Three independent sources converge on the same phenomenon (high AI usage despite low trust) but the specific 66% figure and 31-country attribution do not match any single Ipsos product found. The most parsimonious explanation is that the claim author conflated the KPMG 66% figure with the Ipsos brand and survey methodology.
Gaps¶
| Missing Evidence | Impact on Assessment |
|---|---|
| Full text of all Ipsos AI Monitor editions | Medium — possible that an edition not found contains the 66% figure |
| Google/Ipsos 21-country survey details | Low — this is a different country count than claimed |
Researcher Bias Check¶
Declared biases: The researcher's plural voice advocacy creates motivation to cite the "use despite not trusting" paradox because it supports the article's thesis. This may have led to less careful attribution when the narrative fit was compelling.
Influence assessment: Moderate risk. The confirmation bias danger is not in the phenomenon (which is real) but in the attribution chain (which appears to be wrong). The researcher should verify the source attribution before publication.
Cross-References¶
| Entity | ID | File |
|---|---|---|
| Hypotheses | H1, H2, H3 | hypotheses/ |
| Sources | SRC01, SRC02, SRC03 | sources/ |
| ACH Matrix | — | ach-matrix.md |
| Self-Audit | — | self-audit.md |