Industry surveySalesforce, 2024
No single measure would make a majority trust AI more; transparency comes closest at 42%
Positioning & messaging
Factors That Would Increase Customers' Trust in AI
| Factor | Share of customers (%) |
|---|---|
| Transparency into how AI is used | 42% |
| Human validation of outputs | 35% |
| Greater user control over AI usage | 32% |
| Explainability of outputs | 31% |
| Consistently accurate outputs | 30% |
| Built-in protections within AI products | 29% |
| Government oversight | 28% |
| Third-party ethics review | 28% |
| Industry standards | 23% |
| Public access to underlying algorithms/data sources | 22% |
| Solicitation of customer feedback | 17% |
| N/A - I would never trust AI more | 19% |
- Reader Note:
- Figure covers aggregated customers, meaning consumers plus business buyers. The chart lists factors that would increase customers' trust in AI: transparency into how AI is used leads at 42%, followed by human validation of outputs at 35%, greater user control over AI usage at 32%, explainability at 31%, consistently accurate outputs at 30%, built-in protections at 29%, government oversight at 28%, third-party ethics review at 28%, industry standards at 23%, public access to underlying algorithms or data sources at 22%, and solicitation of customer feedback at 17%. Nineteen percent selected “N/A — I would never trust AI more.” Fieldwork ran from July 26, 2024, through August 20, 2024.
- Evidence:
- Self-reported behaviour
- Vendor interest:
- High — This comes from a company with a commercial interest in this market, so it is not a neutral source.
- Sample:
- Double-blind survey of 16,585 third-party panelists worldwide: 15,015 consumers and 1,570 business buyers across 18 countries on six continents.
Good for
A rough sense of size and direction, and deciding what to look at first.
Not for
- Exact numbers, or saying that one thing caused another.
- This comes from a company with a commercial interest in this market, so it is not a neutral source.
- This is an aggregated or average number that covers very different cases. Real results in a particular segment may differ.




