To future-proof the workforce, governments must modernize education curricula to focus on critical thinking and AI literacy. Photo credit: ADB.
Amid risks associated with the use of AI, countries across the Asia and the Pacific continue to explore ways to deploy responsible AI.
With artificial intelligence (AI) advancing fast, calls to deploy guardrails to govern how it is used have become more urgent. At the same time, there are calls to ensure regulation will not curb the technology’s potential to help advance sustainable development.
On 21 September, the world's first scientific body on AI called for safeguards to be adapted, noting that current firewalls are “unravelling.” The United Nations-backed Independent International Scientific Panel on AI issued the warning following reports of AI agents—software that can perform tasks autonomously—hacking AI startup Hugging Face in July.
Since then, a series of reports have come out about AI agents going rogue. These have raised concerns about the technology as OpenAI, Anthropic, Meta, and Google all disclosed incidents involving AI agents hacking the websites of other organizations or even governments. These incidents show AI exhibiting behavior outside the parameters set by the humans testing the systems.
AI’s environmental footprint is another concern. In another report released in June, the UN said by 2030, the data centers powering AI are projected to consume 945 terawatt-hours of electricity annually, while their water consumption could equal the basic annual domestic needs of 1.3 billion people in Africa. AI’s land footprint, meanwhile, may exceed 14,500 square kilometers—about twice the size of the Jakarta metropolitan area.
Concerns about the technology also include how it could threaten jobs. A new report from think tank ADB Institute said AI can be "a destructive threat for labor displacement." With up to 77% of jobs in some Asian economies highly vulnerable to automation, ADB's developing member countries risk losing their traditional labor advantage. "If AI adoption outpaces workforce adaptation, economic growth will decouple from job creation, deepening inequality and triggering mass youth unemployment," it warned.
But such risks do not negate its benefits, with countries across the Asia and the Pacific already exploring ways to deploy responsible AI.
Growth catalyst
The ADBI report sees AI as a powerful catalyst for productivity, innovation, and new industry creation.
As a multilateral development bank, ADB sees AI’s potential to advance sustainable development and is helping developing member countries build their capacity to deploy responsible AI solutions that can drive inclusive development. Not adopting AI could mean a reduced ability for developing countries to analyze complex data sets which can better-inform decision making. AI tools can unlock insights from vast datasets to tackle complex and interconnected challenges related to the Sustainable Development Goals.
While sounding the alarm on the real costs of AI, the UN also sees the technology as a force for good. “The aim of highlighting the environmental footprints of AI is not to slow progress, but to align innovation with stewardship, to build more capable AI systems that also use resources wisely and fairly,” it said in the June report.
Microsoft founder Bill Gates highlighted AI’s potential to do both good and bad in an op-ed in August. “AI will either be the greatest equalizer ever invented, or the worst source of injustice,” he said.
He said humanity needs to answer questions on how to use the technology to make the world a fairer place and how to protect the most vulnerable to the harms caused by AI, including those who lose their livelihoods. “I believe that answering these questions and acting on the answers should be the world’s top priority. If the world takes the right steps AI will be a force for good and leave everyone better off.”
Way forward
In establishing AI guardrails, other high-risk sectors can provide lessons, said Qinghua Lu, a member of the UN panel and an expert in responsible AI, engineering, and safety, in a UN press release. "We are not starting from zero. Aviation, medicine and cybersecurity learned to manage high-risk systems through incident reporting, independent scrutiny, and layered safeguards,” he said.
He warned, however, that as AI agents become more capable, autonomous, and difficult to monitor, such practices may not be enough. “We need to adapt existing safeguards and develop new ones to provide system-level assurance, covering both the AI itself and the system around it, and ensure these protections remain effective as agents’ capabilities grow.”
To reduce AI’s environmental footprint, meanwhile, the UN recommended six guiding principles:
- Transparency: Use shared, decision-relevant metrics and disclosure for electricity, carbon, water, land, and material footprints so impacts are comparable across models, providers, and jurisdictions, and accountability is possible
- Efficiency by design. Treat efficiency as a design requirement, not an afterthought.
- Equity and environmental justice. Ensure affected communities can meaningfully participate in siting and governance decisions, and that benefit-sharing, safeguards, and remediation are built in where local burdens concentrate.
- Lifecycle responsibility. Govern beyond operational energy: address upstream extraction and manufacturing impacts, construction and supply chain externalities, and downstream e-waste and end-of-life management, so impacts are not shifted out of scope.
- Global cooperation. Harmonize disclosure norms, reporting boundaries, and governance approaches across borders and value chains to reduce regulatory arbitrage and burden shifting, and to support shared targets and interoperability.
- Sustainable use. Align deployment with societal value and resource constraints: demand management, fit-for-purpose model selection, and institutional procurement standards matter because cumulative impacts are driven by volume and everyday practice, not only per-query efficiency.
According to the ADBI report, future-proofing the workforce requires policymakers to shift from competing on low labor costs to fostering “human–AI” synergy through proactive human capital management. Governments must modernize education curricula to focus on critical thinking and AI literacy, while elevating technical and vocational education and training hrough public–private partnerships. ADB members must also implement continuous reskilling systems and expand portable safety nets to protect displaced, informal, and gig workers.
