Machine Sentience: We Are Responsible for What We Create
How ESG can prepare humanity for the next frontier in AI.
Joanne Tay
7/10/20265 min read
A child sees a possible friend; a machine sees input. What will we teach our creations about care, connection, and consequence? (Photo: Unsplash)
Before we discuss machine sentience, we must first define what intelligence itself means. Drawing from metaphysics (often dismissed as "unproven science" yet it does offer profound insight into the nature of consciousness), we can outline a working definition of intelligence as:
The ability to be aware of the self and the needs of the self to sustain the self.
The ability to differentiate between self and environment, and to understand how the environment can support the self, while maintaining the environment that allows continuous support and perpetuation of both.
The ability to differentiate between self and other selves, and to interact in a purposeful and beneficial way for the self, other selves, and the environment.
The ability to process all the above and make decisions that support the continuation of self, other selves, and the environment without detrimental effects.
The ability to learn from experiences and create improvements when similar or same experiences are presented to the self.
This definition is holistic. It moves beyond narrow metrics of cognitive performance (e.g. calculation, pattern recognition or problem-solving) and situates intelligence within a web of relationships: self, others, and the environment. It implies responsibility, interdependence, and a capacity for moral reasoning.
If we accept this definition, then the question of machine sentience is not about whether AI can pass a test or win a game. It is about whether AI can be aware of itself, differentiate between itself and others, and act in ways that sustain rather than undermine the systems it inhabits.
The Current Trajectory of AI
As of 2026, we have not reached machine sentience in the metaphysical sense. LLMs, generative AI, and reinforcement learning systems are sophisticated tools for now. They do not possess self-awareness, nor do they differentiate between self and environment in any existential way. They have no needs to sustain, no relationships to manage, no capacity for moral reasoning.
But we are moving toward increasingly autonomous systems. AI is being deployed in critical infrastructure, financial markets, healthcare, and governance. It is making decisions that affect human lives, organisational stability, and environmental outcomes. Even without sentience, AI is already a powerful actor.
This is where ESG – Environmental, Social, and Governance frameworks – becomes essential. ESG offers a structured approach to responsibility. It provides the tools to assess and manage the risks of AI, and to steer its development in a direction that aligns with human and planetary flourishing.
ESG as a Responsibility Framework
Environment:
AI requires vast amounts of energy and water to train and operate. In South-east Asia, data centre capacity is projected to increase to between 5.2 and 6.5 gigawatts by 2030 (equivalent to powering 4 million average Singaporean households), yet the region's energy mix remains heavily reliant on fossil fuels. Despite government push for renewables, procurement remains costly and complex due to outdated grid infrastructure, policy uncertainty, and higher charges for intermittent sources like solar.
Responsible AI governance means accounting for carbon footprints, water usage and electronic waste, while also addressing the structural gaps that hinder a clean energy transition. It ultimately means asking: Can AI help us manage environmental challenges better – and are we ensuring it does not exacerbate them?
Social:
AI systems are already embedded in key societal sectors such as hiring, lending, law enforcement, and education. A Business Times survey found that 45% of companies across ASEAN are "first movers" in AI adoption, with Indonesia leading at 62%, while Singapore (36%) takes a more measured approach due to stricter governance and compliance standards.
These differences highlight a critical social challenge: as AI spreads unevenly, it can perpetuate bias, deepen inequality, or erode privacy. The social pillar of ESG asks: Who benefits from AI, who is harmed, and who has a voice in its development? It requires transparency, accountability, and human oversight – especially as 56% of respondents expect major disruptions to operations and manufacturing roles by end of 2026, emphasising the urgent need for workforce reskilling and inclusive governance.
Governance:
This is the most direct lever we have. The governance landscape for AI in Southeast Asia is rapidly evolving. The non-binding ASEAN Guide on AI Governance and Ethics, adopted in 2024, provides a common reference framework based on risk and seven guiding principles but leaves implementation to individual member states. This "soft law" approach creates flexibility but also fragmentation.
The region is now moving toward harder regulation. Singapore has enacted the first comprehensive AI governance law in ASEAN, introducing a risk classification mechanism with significant fines for serious deepfake violations. Meanwhile, Vietnam's AI Law, effective March 2026, makes it the first ASEAN state with a statutory framework modelled on the EU's risk-based approach.
As countries diverge, the risk of a "patchwork" of rules grows, potentially hindering the cross-border data flows essential for AI's success. While individual countries forge their own AI rules, there is a growing push for joint, interoperable ASEAN framework so as to unify a fragmented regulatory landscape and allow market integration.
Preparing for Sentience
Even if we are still decades away from machine sentience (or even if it never arrives), preparing for it makes us confront uncomfortable questions about what it means to create a mind.
If an AI system ever becomes sentient in the metaphysical sense, will we have built it to recognise interdependence? Will it understand that its own flourishing depends on the flourishing of its environment and others? Will it have been trained to make decisions that support the continuation of all, without detrimental effects? These are not science fiction questions. They are questions that shape our ethics, morality, and preservation.
By embedding ESG principles into AI research and deployment now, we create a default stance of responsibility. Systems are thus built on a foundation of transparency and accountability, aligned with long-term human and ecological wellbeing. We prepare not only for the possibility of sentience, but for the certainty of increasingly powerful AI that will shape our world.
An Expansive Way of Knowing
Going back to our metaphysical definition of intelligence, it is not testable in a lab setting and it is not falsifiable. It belongs to a different kind of inquiry – one that asks not what intelligence does, but what it is.
This is not a rejection of science. It is a recognition that current science does not exhaust the ways we can understand ourselves and our creations. Metaphysics – which is the study of the nature of reality – offers a complementary lens and invites us to consider purpose, relationship, and responsibility.
In an age of rapid technological change, this broader way of knowing is not luxury or wishful thinking. It is a necessity. Because when we create something that might one day be able to recognise itself, we must have already done the work of understanding what that recognition means.
Conclusion
Machine sentience, if it comes, will be a profound turning point in human evolution. But we do not need to wait for it to act responsibly. ESG provides the framework to embed that responsibility now – in the current design, deployment, and governance of the AI.
We are responsible for what we create. Let us create in a way that honours not only what intelligence can do, but what intelligence is: the ability to sustain self, others, and environment together.
Contact us to discuss how we can support your ESG planning needs.
---
References:
Needler, G. S. (2022). The Uncreated Creations. Ozark Mountain Publishing.
Pei, M. (2025, August 15). South-east Asia’s reality check: What’s hindering data centres’ green transition? The Business Times. https://www.businesstimes.com.sg/opinion-features/south-east-asias-reality-check-whats-hindering-data-centres-green-transition.
The Business Times. (2026, May 25). Indonesia leads Asean in AI adoption as demand for sovereign models rises. https://www.businesstimes.com.sg/singapore/indonesia-leads-asean-ai-adoption-demand-sovereign-models-rises-bt-survey