The business world is changing at two speeds, and the difference between them is widening. The conversation around artificial intelligence (AI) among a growing number of CEOs and boards of directors has changed rapidly over the past year. What was once considered a technical tool is now a strategic priority. Issues that used to concern the chief technology officer (CTO) are now issues for the board of directors. The question for business leaders is no longer whether AI will reshape their industries. It’s about how you position yourself on the right side of that change, before your competitors do.
And there are others. More than 80% of organizations around the world are already using AI in at least one business function, but a global survey of directors found that 66% of boards report having limited or no knowledge of AI, and nearly one-third say it’s not even on their board agenda. In 2024, only 39% of Fortune 100 companies disclosed some form of board oversight of AI, and only 15% of boards received any AI-related metrics. As a result, the governance gap is quietly widening. In Malaysia, the deferred costs of this gap are accumulating.
In many boardrooms, the old assumption that “AI is an IT problem” still persists. The procurement is approved, the CTO installs the system, and the team is briefed. The work is done. But that assumption is outdated. AI systems are already shaping core decisions, such as who is shortlisted for a job, which customers receive which credit terms, how suppliers are ranked, and what information employees see first. These are at the heart of the question of how organizations build trust and how they lose it.
The business case is becoming clearer. According to a 2025 MIT study, organizations with AI-savvy boards outperform their peers by 10.9 percentage points in return on equity. Weak governance has real costs. What is missing is a practical way for boards to approach this issue. Structured, repeatable, and based on real-world business impact. Tools that force the right questions and generate scores that leaders can track and improve.
The ProSocial AI Index is designed to fill that gap, and Malaysian businesses have an opportunity to adopt the ProSocial AI Index early, before external pressures from regulation or reputation force the issue.
the risk that creeps up on you
Let’s consider some situations.
Banks in the Klang Valley use credit scoring tools built primarily on American consumer data. On paper, it performs well enough. In practice, it struggles to value and effectively exclude young semi-urban Malay women. Problems only surface when journalists notice the pattern.
An e-commerce platform deploys a customer service AI bot that is fluent in English and also speaks Malay, but struggles with Chinese and Tamil, the languages that a third of its users actually use. Customer churn is on the rise, and market conditions are believed to be the cause.
A logistics company automates supplier evaluation using a model that does not include ecological variables. We then try to present our sustainability credentials to our institutional partners. Partners ask pointed questions about how environmental impact was measured.
What unites these examples is that there is no routine process to ask what these systems are actually doing, for whom, and at whose expense. Across Southeast Asia, most AI governance frameworks remain voluntary, requiring organizations to take responsibility for themselves.
Prosocial AI Index
The architecture of the ProSocial AI Index is built on a simple structure.
On one axis lies the 4T framework: Customization, Training, Testing, and Targeting. These four aspects determine whether an AI system truly fits its context and purpose.
On the other axis, there is a four-fold bottom line: purpose (does the system enhance human agency?), people (are we fair and honest about who it serves?), profit (are we appropriately distributing economic value within the local economy?), and the planet (is the ecological footprint honestly considered?).
These four T’s and four P’s create 16 rating points. Each is scored using a traffic light scale. Green gets 2 points, yellow gets 1 point, and red gets 0 points. Maximum score: 32.
The results are numbers that executives can track over time, benchmark against peers, and present to boards that want evidence, not narrative. One rule fixes the system. A red score on a target pillar automatically triggers an independent review, regardless of the overall score. Technical elegance cannot compensate for a system that quietly concentrates values or weakens human judgment.
Each pillar addresses a different question.
Tailored asks whether the system fits the context, including language, economic reality, and community.
Trained tracks data to see whose knowledge shaped the model, who was left out, and what the environmental cost of building the model was.
Testing should independently verify real-world impacts, including whether the technology erodes human judgment and autonomous decision-making over time.
Targeted asks the most difficult question: Does this system empower the people who use it, or does it create dependencies that primarily benefit the platform?
What does this do within your organization?
Adopting this index as an internal standard is also a good practice in stakeholder management. Employees who work with AI tools every day often have concerns that are rarely clearly communicated to senior leaders. Will this make my expertise unnecessary? Who will be held responsible if something goes wrong with them?
The index creates a structured way to surface these questions. For HR teams, it provides evidence about the impact on employees rather than providing reassurance. For legal and compliance teams, it is closely aligned with global rules such as EU AI laws that are already impacting global supply chains. Malaysia’s own AI governance framework is largely non-binding. This means that companies that build robust internal standards today will be better prepared to respond should regulations tighten.
For sustainability teams, the index offers another benefit. This helps separate true ESG efforts from the growing problem of AI cleaning. Institutional investors are scrutinizing it more closely with each reporting cycle.
where to start
Start with the three AI systems in your organization that impact most people: customers, employees, and suppliers. Score against the index’s 16 metrics. Share the results internally.
Next, incorporate the four pillars into your vendor evaluation criteria. It is clear that a supplier who cannot answer the targeted pillar questions is sending a signal before the contract is signed.
Set a target score for the next 12 months and assign clear responsibility for improving it.
Small, documented steps can reveal risks that would otherwise be invisible. Organizations that measure these impacts early can address them without delay. The opportunities for Malaysian companies to take the lead in this sector are real and open. The question is whether they accept it now or wait for someone else to define their standards.
Dr. Cornelia C. Walther is a humanitarian worker with over 20 years of experience working for the United Nations. She is an associate professor at the Sunway Institute for Strategic and Competitiveness and a senior fellow at the Sunway Center for Planetary Health at Wharton and the Harvard University Institute for Learning Innovation, with a focus on hybrid intelligence and prosocial AI. She advises UNFPA and the European Policy Center and promotes human agency in the age of AI through the Global POZE Alliance.
