Salesforce’s new AI Trust Quotient shows that AI has a trust problem. Nearly half (48%) of workers in Singapore say it is difficult to get what they want out of AI, and 40% do not trust the data used to train AI systems.
The study, which surveyed 545 full-time workers in Singapore and nearly 6,000 globally, also found that more than half of workers in Singapore (58%) fear humans will lose control of AI, and 94% do not currently trust AI to operate without human oversight.
The trust gap in AI is hindering adoption. Of those who do not trust AI, 95% say they are hesitant to adopt it.
Without accurate data input and humans at the helm, this trust gap will continue growing, preventing businesses from reaping the full benefits of AI.
Key findings
1. AI has a trust problem
Workers find it hard to trust AI even as businesses race to invest in and capitalise on AI’s benefits. This trust gap presents a significant barrier for businesses in adopting and utilising AI effectively.
- Data will make or break workers’ trust in AI: Nearly half (48%) of workers in Singapore find it hard to get what they want from AI right now, and 40% do not trust the data used to train AI systems they see today.
- Fear that humans will lose control of AI: With the growing sophistication of AI, 58% of workers in Singapore fear humans will lose control of AI. 94% do not currently trust AI to operate without human oversight.
- Lack of trust hinders AI adoption: 95% of those who don’t trust AI overall are hesitant to adopt it, and two in three (66%) of those who don’t trust AI training data are hesitant to adopt it.
2. Businesses must tackle the data problem if they want trusted AI to become a reality.
According to Salesforce’s AI Trust Quotient, AI also has a data problem. Without the right data, businesses risk exacerbating the trust gap between workers and AI.
- Lack of complete and accurate data impacting trust in AI: 70% of workers who don’t trust AI say it lacks the information needed to be useful. 65% of workers in Singapore say out-of-date public data and incomplete customer or company data (61%) would break their trust in AI.
- Tech improvements can strengthen trust in AI: To overcome this trust gap, workers in Singapore ranked the following as most important:
- Accurate data (the AI tool uses accurate data) – 84%
- Secure data (the AI tool does not put confidential data at risk) – 82%
- Holistic/complete data (the AI tool uses all possible, relevant data) – 79%
The proof is in the results: 80% of workers in Singapore say that AI needs to consistently produce accurate outputs for them to trust it. This is higher than the global average of 68%. In Australia and India, the figure stands at 73% and 71%, respectively.
Businesses need to break down their disconnected data silos and ensure AI is grounded in a company’s own data.
3. As AI becomes more sophisticated, we must keep humans at the helm.
To unlock the full potential of AI, workers agree that humans must be in the driver’s seat of AI. This combines the best of human and machine intelligence to create more productive businesses, empowered employees, and trustworthy AI.
- Workers in Singapore stress the need for humans at the helm: 94% do not trust AI to operate without human oversight. This is slightly higher than the global average of 90%. This figure stands at 89% and 80% in Australia and India, respectively.
- Bringing AI and humans together: 90% of workers in Singapore do not trust AI to keep data safe on its own, but 59% trust AI and humans to keep data safe together.
- Human-led factors that can drive more trust in AI are: more skill-building or training opportunities (74%), more accessibility and inclusivity (66%)
“Adoption of AI within the workforce is critical if businesses are to achieve their goals of increasing employee engagement and productivity which is foundational to higher value customer relationships and margins,” said Sujith Abraham, Senior Vice President and General Manager, Salesforce ASEAN.
“But in order to use it, they need to trust it. AI is only as good as the data powering it, and new research shows that data makes or breaks the workforce’s trust in AI. Businesses need to unify their data across systems for AI to deliver useful, accurate outputs that workers trust.
“This has to be supported by keeping humans in the driver’s seat of AI, empowering them to focus on the most important outcomes as we enter a new era of AI innovation. Only then can businesses achieve value from AI through better adoption.”
“The findings of Salesforce’s AI Trust Quotient resonate strongly with our experience at AI Singapore. Building trust in AI is crucial for successful adoption, and this requires a multi-faceted approach,” said Laurence Liew, Director, AI Innovation, AI Singapore.
“By prioritising data quality, transparency, and human oversight, we can foster greater trust in AI and unlock its transformative potential for businesses and society.”
