Can we really trust AI, or are we trusting it simply because its answers sound convincing?

In Episode 59 of FAVORIOT Sembang Santai, Zura Huzali and Dr. Mazlan Abbas explore AI trust from a practical perspective.

Using a simple IoT analogy, they ask: if we would not trust an uncalibrated sensor, why should we automatically trust AI that has not been properly tested? 

The conversation covers AI accuracy, bias, privacy, security, Bahasa Melayu and ASEAN language performance, high-risk AI decisions, and why testing needs to reflect the consequences when AI gets something wrong.

Dr. Mazlan also proposes an ASEAN-Korea Trusted AI Sandbox, where real AI applications could be tested across different ASEAN languages, cultures and use cases.

The message is simple:

Test it. Understand it. Then decide where to trust it.

Because perhaps the question is no longer just “Can we trust AI?”

It is “Show me how you tested it.”