AI adoption grows amid falling trust in AI outputs

A recent study by Avanade found that 96% of organisations in the Asia-Pacific region already use artificial intelligence (AI) to support decisions, but trust in AI outputs has declined from 48% in 2023 to just 26% in 2024.

Bhavya Kapoor, president for Asia-Pacific at the Microsoft-focused technology consultancy, noted that the falling trust in AI is an indicator of a more knowledgeable and mature market, adding that the initial frenzy of piloting AI is giving way to a deeper understanding of its complexities, leading to what he terms a “trust deficit”.

This deficit, Kapoor explained in an interview with Computer Weekly, isn’t rooted in a rejection of the technology itself. Instead, it stems from a greater awareness of the inherent challenges that early adopters are now confronting.

“Now, people are more aware of risk and governance issues and that AI is generating information which may not necessarily be accurate if the models are not trained properly,” Kapoor said. “A lot of organisations started early with pilots without thinking about all of that and that’s what’s causing the trust deficit.”

This observation comes as many organisations across the region find themselves in the early stages of AI adoption. While enthusiasm and budgets for AI are growing, Avanade’s study found that nearly half are still in pilot or proof-of-concept phases, grappling with the next steps.

Kapoor recounted a recent meeting in Singapore with more than 60 C-suite executives where nearly all had used generative AI. However, fewer than 10% had thoroughly considered the strategic implications of governance, scalability and corporate culture. He drew a parallel to the early days of e-commerce and ride-sharing apps, when users were hesitant to save their payment information in those apps.

“When we first started using Uber, I don’t think any of us stored our credit card numbers with them,” he said. “But now, you have your credit cards stored there for years. It’s a matter of time before people get more comfortable.”

To address the trust deficit, organisations will need to move from isolated AI experiments to building robust, enterprise-wide foundations. This includes creating clean, reliable data platforms and establishing strong governance frameworks – areas where Avanade is advising clients to adopt a long-term perspective to prevent the siloed chaos that has affected data architecture projects.

A direct result of this heightened risk awareness is also the renewed and urgent focus on cyber security. Kapoor noted that the widespread implementation of generative AI has significantly expanded the threat surface, forcing businesses across the region to address a long-standing weakness.

“I would say over decades, Asia businesses have underinvested in cyber security,” Kapoor said. “Even if you have one AI use case to pilot, your entire organisation’s data is there so it’s important to have guardrails.”

For Avanade, AI and cyber security present significant business opportunities, particularly in the mid-market, an underserved segment which the company defines as businesses with revenues of between $1bn and $7bn.

Kapoor noted that mid-market businesses are often more agile and bold, but typically operate with smaller technology teams and thus require a partner that can deliver a full spectrum of technology services, from design, build, and operations to security.

At the same time, Avanade is doubling down on key markets in the region, specifically Indonesia, India and Japan. “We believe Indonesia’s potential is still under-tapped, and we’re making significant investments there,” Kapoor said, noting the recent launch of a local Microsoft datacentre as a key enabler.

In Japan, Avanade is seeing a unique driver for AI adoption. “Japanese people are not native English speakers, so a lot of translation is done by Copilot,” Kapoor said, adding that this high-value use case is driving widespread AI adoption among Avanade’s clients and employees in the country.