Moderator and Panellists
Panellists
- Lauren Perry, Responsible Technology Policy Specialist, Human Technology Institute, University of Technology Sydney, Australia
- Dr Nopparuj Chindasombatcharoen, Research Fellow, Thailand Development Research Institute (TDRI), Thailand
- Panharith In, Head of Innovation, Cambodia Academy of Digital Technology (CADT), Cambodia
- Yohanes Lukiman, Board Member, ASEAN-BAC Indonesia (unable to attend in person; perspective shared by the moderator)
Moderator and facilitators
- Arifah Sharifuddin, Institute Director, Tech for Good Institute (TFGI)
- Fairoz Ahmad, Programme Fellow, Tech for Good Institute (TFGI)
- Erlanggasakti Ubaszti Putra, Programme Analyst, Tech for Good Institute (TFGI)
Key Takeaways
1. Distrust of AI Has More Than One Root
Panellists offered a more nuanced reading of why SMEs hesitate to adopt AI. In Cambodia, most SMEs still equate AI narrowly with generative tools for marketing content, with little sense of how it could improve the rest of the business, and imported AI tools trained on foreign data can feel unreliable in a local context. One panellist added a further view: that some of what looks like distrust may be risk aversion or a gap in understanding AI’s benefits, since firms rarely apply the same scrutiny to other opaque technologies such as cloud migration. The discussion pointed to a structural response rather than leaving each SME to work it out alone: an ecosystem of vetted service providers, which lowers search costs and shifts accountability toward the vendor, with government setting the regulatory foundation and industry building the day-to-day guardrails.
2. Inclusion and Trust Are Built Locally, Not Assumed
Panellists warned against treating infrastructure and programme design as separate problems. Experience from the Pacific Islands shows that cultural mistrust of AI is often linked to a high prevalence of financial scams, a pattern also seen in parts of Southeast Asia. The ITU and UN’s Smart Villages and Smart Islands programme, which pairs infrastructure rollout with community-based digital hubs, was cited as a model for building local ownership and trust together. Indonesia’s QRIS was the clearest regional proof point: designed around the smartphones and agent networks the country already had, it reached 42 million merchants, nine in ten of them MSMEs, across 13.6 billion transactions in 2025. The panel argued that AI skilling delivery should borrow the same discipline: local language, peer-led, and low-bandwidth by default.
3. Skilling and Financing Must Meet SMEs Where They Are
Both TFGI’s initial consultation and the panel discussion converged on the same diagnosis: most current AI skilling for SMEs teaches a tool rather than matching training to where a business actually sits on its adoption journey. One panellist pointed to high-return, sector-specific use cases, such as AI cameras for quality control in manufacturing that pay back in under a year, and stressed that training needs executive buy-in, not just junior staff. Panellists endorsed a matching, graduated financing model: grants for early-stage SMEs to test small use cases, matched investment as they grow, and commercial financing once a use case and return are proven. Guidance also needs to fit time-poor SME leaders. Drawing on the Human Technology Institute’s Safe AI Adoption for SMEs work, one panellist noted that safety guidance works best when it is fast, practical, and focused on real-world harms such as bias, discrimination, and privacy risks, rather than on abstract, developer-level debates.
What Happens Next
The insights and recommendations from this discussion will feed into the ATTS Policy Memorandum, which will be submitted to the ASEAN Secretariat and the Permanent Representatives of ASEAN Member States ahead of the ASEAN Summit in November. We thank the Singapore Institute of International Affairs (SIIA) for convening and coordinating the summit, the ASEAN Secretariat for its support, and our moderator and panellists for a rich discussion.
