\n\n\n\n
The numbers are not ambiguous.<\/strong> McKinsey and BCG consistently find that 70% of digital transformation initiatives fail to meet their stated objectives. Bain’s 2024 analysis puts the figure for broader business transformations at 88%. Globally, an estimated $2.3 trillion has been wasted on unsuccessful programs \u2014 with $3.4 trillion more projected to be spent by 2026 at the same failure rate. This is not a technology problem. The UNDP said as much in a report published this month: governments and institutions do not lack digital ambition \u2014 they lack the institutional conditions for that ambition to take root.<\/em><\/p>\n\n\n\n That gap between ambition and conditions is where trust gets misplaced.<\/p>\n\n\n\n Institutions tend to trust in three directions, all of them wrong.<\/strong> They trust vendors \u2014 because the vendor is certified, proven, and backed by a reference list. They trust compliance \u2014 because the ISO certificate is on the wall, the audit passed, and the checklist is complete. And they trust the roadmap \u2014 because someone spent three months building it, and it has phases, milestones, and a go-live date.<\/p>\n\n\n\n None of these are trust. They are proxies for trust. They replace verification with documentation. And when the system fails \u2014 as the UK’s National Health Service discovered after spending over \u00a39.8 billion on a national electronic health records program that was cancelled in 2011 having delivered just 2% of its promised benefits \u2014 no one can answer the most basic question: who was responsible for making sure this worked?<\/em> The vendor blamed the specifications. The department blamed the vendor. Parliament blamed the programme. The money was gone, and the accountability had been distributed so thoroughly that it had effectively ceased to exist.<\/p>\n\n\n\n The NHS case is not exceptional. It is the template.<\/p>\n\n\n\n Indonesia’s own story sits within this same pattern, at a different scale.<\/strong> By 2023, Indonesian government institutions were running 27,000 applications across ministries, agencies, and regional offices \u2014 each built separately, each consuming budget, each operating in its own silo. The annual operational cost: Rp 8.1 trillion for data centres alone, Rp 2.7 trillion for application operations. As Kominfo itself acknowledged, interoperability between these systems was so poor that building data-driven policy had become structurally impossible.<\/p>\n\n\n\n This did not happen because of incompetence. It happened because the wrong thing was being measured. When leadership does not have a clear framework for evaluating digital systems \u2014 when technology is unfamiliar territory \u2014 the temptation is to measure what is visible: the number of applications built, systems deployed, projects completed. The application becomes the KPI. The problem the application was supposed to solve gets forgotten. Digitisation is treated like buying an iPhone \u2014 visible, announcceable, prestigious \u2014 rather than what it actually is: an investment that must be justified by the efficiency and capability it creates.<\/p>\n\n\n\n This is not a technology literacy problem. It is a leadership maturity problem. Civil servants do not need to become engineers. But they need to know the right questions to ask. The most dangerous institutional failure is not the leader who admits they do not understand something. It is the leader who does not know they are lost.<\/p>\n\n\n\n Indonesia’s national SPBE index reached 3.12 out of 5 in 2024, rated “Good” \u2014 and in the same year, the Pusat Data Nasional was brought down by a ransomware attack that encrypted critical government systems, disrupted immigration checkpoints, BPJS, and dozens of public services, and exposed the absence of a basic safeguard: there was no data backup. The SPBE score measures maturity of implementation. It does not measure whether the trust architecture actually holds under pressure. Those are two different things. A compliance score is not a resilience score.<\/p>\n\n\n\n Now add AI into this picture.<\/strong> Institutions that have not resolved the foundational question of trust are now deploying artificial intelligence at scale. The global enterprise adoption of AI grew 115% between 2023 and 2024. Yet only 40% of organisations have deployed basic safeguards \u2014 explainability mechanisms, audit trails, ethical frameworks. Nearly half admit to a “trust gap” between what AI promises and what it delivers.<\/p>\n\n\n\n This matters in a way that is structurally different from previous technology failures. When a conventional system fails, there is usually a log, an audit trail, a point where the error can be identified and attributed. When an AI system produces a wrong outcome, the question of why<\/em> often cannot be answered \u2014 not because the answer is hidden, but because the system was not designed to provide one.<\/p>\n\n\n\n The consequences of this opacity are already visible. UnitedHealth’s subsidiary deployed an algorithm called nH Predict to assist in determining the length of post-acute care for Medicare patients. A lawsuit filed in 2023 alleges the system had a 90% error rate \u2014 and that the company continued using it because fewer than 0.2% of patients appealed their denials. When the algorithm’s decisions were appealed, more than 90% of those appeals were overturned. The company’s response: the algorithm was a decision-support tool, not a decision-making tool. The actual decisions were made by licensed physicians. Those physicians said they had little practical choice but to follow the system’s output.<\/p>\n\n\n\n There is the accountability gap in its most modern form. Not a missing backup. Not a failed vendor. An institution that deployed a system it could not explain, into a workflow it could not audit, producing outcomes it could not verify \u2014 and when the outcomes were questioned, no one could answer for them.<\/p>\n\n\n\n AI does not create this problem. It inherits it and amplifies it. Institutions that delegated trust before \u2014 to vendors, to compliance, to roadmaps \u2014 will delegate it again, this time to models. And the consequences will be faster, wider, and harder to trace.<\/p>\n\n\n\n The rules of the game are genuinely different with AI. Conventional systems are deterministic: the same input produces the same output, and the logic can be inspected. AI systems are probabilistic: outputs vary, reasoning is often opaque, and errors may not surface until they have already propagated through thousands of decisions. An institution that treats an AI deployment like a software procurement \u2014 select vendor, sign contract, go live \u2014 is not just making a technology mistake. It is building on a foundation that was already cracked and adding a faster, less transparent layer on top of it.<\/p>\n\n\n\n What needs to change is not the technology. It is the question institutions ask before any of this begins.<\/strong> The question is not: which vendor should we trust?<\/em> It is: how will we verify that this system is doing what we think it is doing, and who is accountable for that verification?<\/em><\/p>\n\n\n\n Trust as architecture means building systems where accountability is designed in, not assumed. It means measuring outcomes \u2014 problems solved, services improved, errors caught \u2014 not outputs: systems deployed, applications launched, roadmaps completed. It means treating digital investment the way serious organisations treat any critical infrastructure: with maintenance budgets, resilience requirements, and someone who can be named when something goes wrong.<\/p>\n\n\n\n And it means that before any institution deploys AI, they must be honest about what they do and do not understand about how it works \u2014 not to avoid adoption, but to earn the right to it. The most dangerous place an institution can be is not behind on technology. It is confidently moving in the wrong direction, unaware that the map they are following was never designed for the terrain they are crossing.<\/p>\n\n\n\n Every major institutional digital transformation begins with the same ritual. A vendor is selected \u2014 one with an impressive client list and a reassuring deck. A roadmap is signed. Certifications are exchanged. Leadership announces the initiative with ambition. And then, somewhere between the third phase and the go-live date, something breaks. Not because the technology […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[18],"tags":[],"class_list":["post-252","post","type-post","status-publish","format-standard","hentry","category-cyber-trust"],"_links":{"self":[{"href":"https:\/\/desyanta.com\/index.php?rest_route=\/wp\/v2\/posts\/252","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/desyanta.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/desyanta.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/desyanta.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/desyanta.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=252"}],"version-history":[{"count":1,"href":"https:\/\/desyanta.com\/index.php?rest_route=\/wp\/v2\/posts\/252\/revisions"}],"predecessor-version":[{"id":254,"href":"https:\/\/desyanta.com\/index.php?rest_route=\/wp\/v2\/posts\/252\/revisions\/254"}],"wp:attachment":[{"href":"https:\/\/desyanta.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=252"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/desyanta.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=252"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/desyanta.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=252"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
\n\n\n\n
\n\n\n\n
\n\n\n\n
\n\n\n\n![\"\"](\"https:\/\/desyanta.com\/wp-content\/uploads\/2026\/05\/misplaced_trust_infographic-450x1024.png\")
<\/figure>\n","protected":false},"excerpt":{"rendered":"