A Singapore Government Agency Website
How to identify
The gap between AI course completion and real-world application is leaving training investments without impact
Approximately 70,000 Singapore citizens take up SSG-funded AI courses each year. Yet despite completing formal training, many are likely not translating their learning into practice. A study on M365 Copilot adoption found that while 9 in 10 learners acknowledged the value of formal training, 7 in 10 bypassed structured onboarding entirely, preferring to learn through trial and error or peer discussion instead. Applied to the Singapore context, this suggests that a significant proportion of the 70,000 learners may be leaving courses with theoretical knowledge but without the hands-on experimentation and peer support structures needed to apply AI confidently in real life.
Who they are Singapore citizens aged 18 and above who have completed SSG-funded AI courses, spanning working adults across various industries, freelancers, and other members of the adult population.
What they are trying to do They want to use AI tools to improve their productivity and simplify tasks in their daily work or personal lives — such as drafting documents, analysing information, automating repetitive tasks, or solving problems more efficiently. Some are also motivated to go further, exploring more advanced applications such as Retrieval-Augmented Generation (RAG), building AI-powered applications, or agentic AI workflows.
What task they are trying to complete After completing their AI course, they attempt to independently apply what they have learned to real work or life tasks. However, they struggle to bridge the gap between the structured course environment and the unstructured reality of applying AI on their own. For those seeking to go beyond basic use, they quickly encounter a practical barrier — more advanced AI applications require paid subscriptions to access higher token limits or more capable models, which free tiers do not support.
Where and when they encounter this problem The problem surfaces immediately after course completion, when learners return to their workplace or daily routines without guided support. Learners without the financial means to upgrade their access are effectively capped at superficial usage, discouraging deeper skill development. Over time, this creates a two-tier outcome — those who can afford to pay progress, while those who cannot disengage or plateau, leaving the training investment unrealised for a significant portion of the 70,000 learners each year.
Misalignment between course design and how adults actually learn Formal AI courses deliver structured, theoretical content, but research suggests most learners prefer experiential learning (trial and error) and social learning (peer discussion) over structured onboarding. Courses may not be designed to accommodate these preferences, leaving learners underprepared to self-direct their learning after graduation.
Absence of post-course support structures Once the course ends, learners are left to apply their skills independently without guided practice, peer communities, or mentorship. The structured environment that supported learning during the course disappears precisely when learners need it most — during the critical transition to real-world application.
Confidence gap Learners may understand AI concepts theoretically but lack the confidence to experiment and apply them in their actual work context. Without low-stakes environments to practise and make mistakes, confidence does not develop into competence.
Access and affordability barriers Free AI tool tiers impose token limits and restrict access to more advanced models, preventing learners from exploring higher-value applications such as RAG, agentic AI, or building AI apps. Learners whose companies do not sponsor AI subscriptions are structurally capped at superficial usage, regardless of their motivation or ability, resulting in a lack of reinforcement of learning. To address this, PM Wong announced at Budget 2026 that premium AI subscriptions will be given free upon completion of qualifying courses — making tool access an entitlement rather than a privilege.
Lack of contextual relevance during training AI courses may teach generic skills that learners struggle to map onto their specific job roles or daily tasks. Without concrete, role-relevant examples during training, learners leave without a clear mental model of how AI fits into their own lives.
Habit formation is not supported Building AI proficiency requires repeated, consistent practice over time. Without nudges, accountability mechanisms, or structured follow-through after the course, learners revert to familiar workflows and AI usage fades before habits can take root.
Operational and process barriers to scaling the subscription initiative While the free premium subscription addresses the affordability gap, delivering it at scale introduces several process pain points that must be resolved for the initiative to work effectively. These include the need to integrate with multiple AI subscription vendors (such as Google and OpenAI), each with their own distinct redemption and sign-up processes. Additionally, the redemption mechanism must be designed to avoid collecting credit card details, which risks triggering auto-renewal disputes, and must include safeguards to prevent the reselling of redemption codes.
Training investments go unrealised at scale Failing to support learners beyond the classroom wastes not just public training resources, but the very motivation and potential these learners bring. The cost compounds year on year as each new cohort faces the same barriers.
Workers risk being left behind as AI becomes embedded in the workplace Without access to AI tools to reinforce their learning, learners grow less familiar with AI over time, eventually forming a growing pool of workers who are not AI competent. This is increasingly concerning as more companies integrate AI tools into their core workflows. Workers who cannot navigate and use these tools effectively risk becoming less employable and less productive, undermining both their own career prospects and their contribution to the broader workforce and economy.
A two-tier AI proficiency divide widens Without solving the access and affordability barriers, AI proficiency will continue to correlate with financial means. Learners from lower-income backgrounds or whose employers do not sponsor subscriptions will remain capped at superficial AI usage, while those with resources pull further ahead. Over time, this risks entrenching a new form of digital inequality among Singapore citizens, with consequences for social mobility and workforce inclusivity.
Singapore's national AI readiness goals are set back Singapore's broader ambition to build a digitally proficient, AI-ready workforce depends on citizens being able to apply AI meaningfully, not just complete courses. If the gap between learning and application persists across 70,000 learners a year, the cumulative drag on national productivity and AI adoption will be significant, weakening Singapore's competitive position in an increasingly AI-driven global economy.
Learner motivation and trust erode Citizens who invest time completing AI courses but find themselves unable to apply what they have learned — due to confidence gaps, lack of support, or tool access barriers — are likely to disengage from future upskilling efforts. Repeated experiences of learning that does not translate into real value can erode trust in government-led training programmes more broadly, making future initiatives harder to drive adoption for.
To validate the need for this initiative, user research was conducted through a combination of primary and secondary methods. A public survey was administered via CrowdTask, gathering responses from 500 Singapore citizens aged 18 and above, spanning students, working professionals, retirees, and homemakers. The findings were analysed using AI-assisted analysis of the raw survey data. This was complemented by analytics drawn from SSG's data warehouses and a review of publicly available online materials on AI adoption trends.
The survey found that the majority of respondents already use AI in some form, with ChatGPT, Gemini, and Copilot being the most widely used tools. A smaller but notable segment of roughly 20–25% reported not using AI at all, skewing slightly older, citing reasons such as not knowing how to start, lack of time to learn, or not seeing a need yet. Common use cases included drafting and refining documents, research and fact-finding, explaining concepts, and coding assistance, with more advanced users exploring automation workflows and agentic AI tasks.
Several pain points were consistently surfaced across demographics. Token and credit limits on free-tier tools were the single most cited frustration, frequently disrupting workflows and discouraging continued use. Cost emerged as a significant barrier, with a large proportion of respondents expressing willingness to use AI but unwillingness to pay for subscriptions. Other recurring challenges included concerns about accuracy and hallucinations, difficulty with prompt engineering, and data privacy and security concerns — particularly among professionals handling sensitive information.
When asked what form of government support would be most valuable, free or subsidised access to AI tool subscriptions was the top ask across all demographics, far outpacing demand for courses. Many respondents explicitly noted that hands-on access to paid tools was more valuable than attending a class. Notably, older respondents showed genuine enthusiasm for AI but faced steeper learning curves and felt underserved by existing offerings, while younger respondents tended to be more self-sufficient but were constrained by cost. These insights directly informed the initiative's approach of provisioning AI subscriptions to eligible learners upon course completion, rather than limiting support to training alone.
Several market and operational risks were identified and mitigated in the design of the initiative.
Demand uncertainty and inventory management.
A key risk in any subscription provisioning initiative is the difficulty of accurately forecasting demand upfront, which can result in over-purchasing unused licenses or under-purchasing and leaving eligible learners without access. To mitigate this, AI subscriptions will be procured in phases, calibrated against actual redemption data and course completion rates from SSG's systems. This phased procurement approach ensures that public funds are deployed efficiently, with each subsequent phase informed by observed demand from the preceding one. GovSupply's inventory management system provides real-time visibility of code stock levels, enabling proactive replenishment before inventory runs low.
Vendor dependency and supply continuity.
The initiative's reliance on AI vendors for the timely supply of redemption codes introduces a risk of supply disruption. This is mitigated through clear service level agreements with vendors on code provisioning timelines and minimum stock thresholds, with escalation procedures in place should supply commitments not be met. The architecture also supports the onboarding of additional AI vendors over time, reducing single-vendor dependency as the initiative scales.
Fraudulent and erroneous redemptions.
The risk of ineligible individuals attempting to redeem subscriptions is mitigated through Singpass authentication at the point of redemption, which verifies the identity and citizenship status of each applicant. Eligibility checks are performed against SSG's enrolment and attendance records, and GovRewards enforces single-entitlement rules to prevent duplicate redemptions. Together, these controls ensure that subscriptions are provisioned only to verified and qualifying learners.
Low redemption uptake.
The risk of eligible learners not redeeming their subscriptions is mitigated by embedding the redemption process within the existing course completion journey, ensuring that learners are informed of their entitlement at the point of eligibility. The use of FormSG as the frontend interface provides a simple and familiar redemption experience that minimises drop-off. Redemption rates will be monitored closely during the pilot phase, with targeted outreach and process improvements implemented if uptake falls below expectations.
Platform and integration risks.
The initiative's dependence on multiple WOG platforms introduces integration complexity and the risk of service disruptions affecting the redemption journey. This is mitigated by leveraging established platforms with proven track records and built-in resiliency features. Manual fallback procedures and standard operating procedures will be documented to ensure operational continuity in the event of platform disruptions, and integration points will be thoroughly tested prior to launch.
The initiative involves close coordination across multiple internal and external stakeholders. Within SSG, the relevant business, data, and IT teams have been engaged to ensure alignment on eligibility criteria, data flows, and system integration requirements. WOG platform product teams, including those responsible for GovRewards, GovSupply, and OPUS, have been consulted to confirm platform suitability and integration feasibility. AI vendors have been engaged on code provisioning arrangements. Stakeholder buy-in has been established across these parties to support the initiative's implementation.
The initiative directly addresses the key pain points surfaced through user research. The single most requested form of government support across the 500 survey respondents was free or subsidised access to AI tool subscriptions, and this initiative delivers exactly that by provisioning premium AI subscriptions to eligible learners upon completion of qualifying courses. By tying access to course completion, the initiative also incentivises learning while ensuring that subscriptions are directed to learners who are motivated to use them. The use of Singpass for identity verification and WOG platforms for eligibility checking addresses concerns around data privacy and fraudulent access, building trust in the redemption process.
The key user flow begins with a learner completing a qualifying AI course under SSG. Upon completion, the learner submits a redemption request via FormSG, selecting their preferred AI tool from a predefined list. The system then orchestrates eligibility checks against SSG's enrolment and attendance records and GovRewards to enforce single-entitlement rules. Once eligibility is confirmed, an available redemption code is retrieved from GovSupply's inventory and issued to the learner. OPUS orchestrates the end-to-end workflow across the various systems, while Postman enables API integrations between them.
The redemption experience is designed to be simple and low-friction for the learner. FormSG serves as the frontend interface, providing a familiar and accessible form-based experience that requires no account creation or technical knowledge. Singpass authentication removes the need for learners to manually declare their personal particulars, streamlining the submission process and reducing the risk of erroneous or fraudulent submissions. The end-to-end redemption journey is designed to be completed promptly upon submission, minimising waiting time. Standard operating procedures and a clear appeals process will be established to support learners who encounter issues during the redemption process.
User validation findings will be incorporated following the pilot launch of the initiative. Feedback gathered during the pilot phase will be used to identify pain points in the redemption journey and inform iterative improvements to the system design and user experience prior to full-scale rollout.
The North Star metric for this initiative is the redemption rate among eligible learners — that is, the proportion of learners who successfully redeem their AI subscription upon completing a qualifying course. This metric reflects both the accessibility of the redemption process and the perceived value of the subscription to learners, and serves as the primary indicator of the initiative's effectiveness in translating course completion into meaningful AI access.
In the short term, the initiative is expected to increase access to premium AI tools among learners who have completed qualifying courses, directly addressing the cost barrier identified as the key obstacle to broader AI adoption. In the medium term, as more learners gain hands-on experience with AI tools, the initiative is expected to contribute to improved AI literacy and confidence across the workforce, supporting SSG's broader mission of equipping Singaporeans with future-ready skills. In the long term, the modular architecture of the initiative allows it to be scaled to accommodate new AI vendors, expanded eligibility criteria, or new course categories, positioning it as a sustainable and adaptable platform for ongoing AI upskilling support.