The AI landscape doesn't move in one direction — it lurches. Some techniques leap from experiment to table stakes in a single quarter; others stall against regulatory walls, technical ceilings, or organisational inertia that no amount of hype can dislodge. Knowing which is which is the hard part. The State of Play cuts through the noise with a rigorously maintained index of AI techniques across every major business domain — classified by maturity, evidenced by real-world adoption, and updated daily so you always know where you stand relative to the field. Stop guessing. Start knowing.
A daily newsletter distilling the past two weeks of movement in a domain or two — delivered to your inbox while the index updates in the background.
Each dot marks the weighted maturity of practices within a domain — hover for a brief summary, click for more detail
Training programmes and certification for employees on responsible AI development, deployment, and use. Includes role-specific responsible AI curricula and assessment; distinct from general AI literacy which covers capabilities rather than governance responsibilities.
Responsible AI training and certification has transitioned from infrastructure scarcity to ecosystem saturation. Supply-side expansion is now relentless: Microsoft systematically retired foundational AI/ML certifications in mid-2026 in favor of agentic-AI and specialized business applications; EC-Council released CRAGE aligned to NIST and ISO standards; Joint Commission (23,000+ healthcare organizations) launched governance-focused certification; Coursera's platform is accelerating enrollments (20 per minute as of June 2026, up 33% YoY); the U.S. government committed $369M to AI Literacy Framework across 56 state hubs; and MIT RAISE reached 24M learners globally. Yet adoption remains constrained by organizational execution, not credential availability. Gallup's 2026 workplace study revealed only 12% of employees strongly agree AI changed how work is done despite rapid deployment—signaling training programs are not translating into behavioral change at scale. Infosys found only 2% of 1,500 large enterprises meet responsible AI standards despite infrastructure maturity. The core barrier is not training availability but training effectiveness: 85% of employees report receiving training that fails to help them apply AI to actual work, and only 4% of organizations deliver role-specific training aligned to business outcomes. Organizations with structured, measurable upskilling move Responsible AI competency from 25% to 81% accomplished (Workera benchmark of 88,000 assessments), and achieve 42% significant ROI versus 21% baseline. However, a critical negative signal persists: the market is producing certificate holders faster than builders—only 1 qualified AI engineer available per 10 open GenAI positions, with credential proliferation decoupling from actual deployment capability. Barriers are organizational—measurement discipline, translating training into workflow redesign, embedding expertise into decision-making—not epistemic. This gap positions responsible AI training as a critical differentiator between organizations that can operationalize AI safely and those that deploy in reactive, high-risk modes, but credential abundance masks profound capability shortages.
The credential and training ecosystem reached saturation in Q1–Q2 2026 with an inflection point at mid-year. Microsoft's systematic retirement of foundational AI/ML certifications (Azure Data Scientist, Azure AI Engineer, AI Fundamentals all retired June 2026) signals vendor pivot from foundational to agentic-AI and domain-specialized competencies. EC-Council launched CRAGE (11 modules, NIST/ISO-aligned) addressing where 95% of AI initiatives fail to reach production. Joint Commission—evaluating 23,000+ healthcare organizations—launched Responsible Use of AI in Healthcare (RUAIH) governance-focused certification, signaling healthcare sector acceleration. RAI Institute's RAISE Pathways progressed to tiered organizational-level badges with automated governance tracking. Coursera platform enrollment in AI courses accelerated to 20 per minute (June 2026, up 33% YoY). MIT RAISE reached 24M learners in 175 countries. The U.S. government operationalized $369M in training infrastructure across 56 state hubs (DOL/NSF TechAccess: AI-Ready America) with standardized 5-competency AI Literacy Framework. Asia-Pacific governments (India NAIDM, Singapore mandating 40,000-person program, Philippines certification standards) operationalized mandatory training as governance infrastructure. By late June 2026, EU AI Act Article 4 compliance mandate (August 2, 2026 effective date) created urgency for documented, role-specific, ongoing AI literacy training with audit-ready competence records.
Deployment signals reveal persistent effectiveness gaps masked by supply-side enthusiasm. Organizations with structured, measured upskilling move Responsible AI competency from 25% to 81% (Workera benchmark of 88,000 assessments), and mature programs report 42% significant ROI versus 21% baseline. Context-specific training works: TVET trainer AI literacy program achieved 70.6% completion and 90% confidence gains; NSW government operationalized role-based training identifying specific capability gaps; UK Skills England launched SKAI framework with 150+ employer partners and role-based apprenticeships. However, a critical capacity signal emerged in June 2026: the market faces acute talent shortage despite training proliferation—only 1 qualified AI engineer per 10 open GenAI positions, with 53% projected shortfall, indicating credential production is decoupling from actual builder/practitioner capability. Implementation remains constrained by execution discipline, not infrastructure. Research identifies the core problem: 85% of employees report training fails to help them apply AI to actual work; only 4% of organizations deliver role-specific training; only 25% of L&D departments fully align training to business strategy; hierarchical access gaps show C-Suite 3.4x more confident than frontline staff despite identical tool deployments. McKinsey/WRITER survey (2,400 execs) shows knowledge and training deficits as #1 barrier to responsible AI deployment. The bottleneck has shifted entirely from supply (resolved) to organizational execution: measurement infrastructure, role-differentiated curriculum, translating training into workflow redesign, and accountability for behavior change and business outcome impact remain the constraining challenge.
— Synthesis of credible research reveals critical negative signal: 82% of orgs provide training yet 59% report skills gaps; root cause identified as structural learning gap (not knowing what to do), with only 5% of orgs achieving substantial AI financial gains.
— 79% of enterprise AI job postings cite AI literacy as requirement (driven by EU AI Act Article 4, effective Aug 2026); gap shows companies lack baseline training, role-specific content, completion records—creating regulatory compliance urgency.
— Microsoft's 5-course professional cert on Coursera (updated July 2026) covers NIST AI RMF, ISO/IEC 42001, GDPR, FAIR with hands-on governance projects, showing major vendor scaling of governance training.
— ANAB-accredited ISO 42001 cert body survey (525 US respondents, April-May 2026) reveals governance maturity gap: 90% funded, 74% claim audit-ready, only 27% actually mature; mature governance correlates with 78% production AI agent deployment.
— Large-scale study (35 leaders, 1,300 employees) shows 55.1% use AI weekly/daily but only 33.3% received training; only 48% have sufficient time/resources; organizations focus on upskilling not reskilling, missing AI-transformation readiness needs.
— UK government guidance (DWP/Skills England) identifying effective AI upskilling approaches: embedded in day-to-day work, practical task-based, structured with clear pathways; system-level training infrastructure too slow to match adoption pace.
— EC-Council's 11-module credential explicitly aligned to NIST AI RMF and ISO/IEC 42001, addressing governance skills gap with structured training; North America-specific offering reflects market demand.
— Peer-reviewed analysis (17 authors) argues responsible AI practices exist internally but produce no external market signal; proposes outcome-oriented, independently-verified certification to close trust gap and enable market differentiation.
2023-H1: Microsoft Research releases Responsible AI Maturity Model with 24 empirically-derived dimensions; Microsoft and LinkedIn launch AI skills certification with responsible AI components. However, survey data shows persistent training gaps—86% of workers need training but only 14% receive it, with large confidence disparities between leadership and frontline staff.
2023-H2: Research documents growing gap between published responsible AI frameworks and practitioner ability to operationalize them; UK government promotes certification as enabler of trustworthy AI; multiple studies confirm persistent disconnects between official guidance and actual organizational practice in training implementation and stakeholder engagement.
2024-Q1: Global survey (1,000 organizations, 20 industries) confirms significant strides in RAI maturity alongside persistent implementation gaps; CHI conference research identifies nine critical training topics needed by knowledge workers but reports inadequate employer support; Workday survey shows 80% of organisations lack published guidelines on responsible AI use; UK government publishes Responsible AI Toolkit (March 2024) with structured assurance and implementation resources. Market transitions from "what frameworks exist" to "how do we operationalise them," but execution gap widens as deployment velocity accelerates.
2024-Q2: Professional bodies accelerate certification development (GARP Risk and AI, ISACA Audit Toolkit); Microsoft publishes inaugural Responsible AI Transparency Report detailing governance maturity; third-party analysis confirms maturity gap persists—only 20% of companies report mature RAI programs, 30% have none. Academic research begins systematic evaluation of RAI implementation effectiveness. Concerns emerge around shadow AI and reactive organizational posture despite regulatory catalysts (EU AI Act). Training infrastructure investment accelerates but remains concentrated among large vendors and early-adopter organizations.
2024-Q3: Supply-side infrastructure accelerates with U.S. government (GSA/OMB) launching structured AI training series and ISACA releasing ethics/audit curriculum; institutional adoption signals maturation. Demand-side paradox deepens: 97% of AI leaders commit to responsible AI but 48% lack resources; EY survey reveals only 37% of U.S. leaders upskill employees "fully at scale" despite 95% investing in AI. Academic research pivots to ROI justification frameworks, suggesting execution barriers are now organisational (resource allocation, cost models) rather than epistemic (not knowing what to do).
2024-Q4: Supply-side training continues expanding with government and vendor infrastructure deployments. However, new data in October-December reveals the persistence of execution gaps at scale: Precisely/Drexel research shows 60% of organisations cite lack of AI skills/training as barrier to AI initiatives; Stibo Systems survey of 500+ U.S. business leaders documents 58% lack AI ethics training despite 86% wanting it, and 49% unprepared for responsible AI use. Stanford research reveals that even equipped teams struggle with practical application—AI product teams unable to execute fairness evaluations due to knowledge gaps, and developers continue learning via self-study rather than structured training. Pattern holds: training infrastructure and commitment exist, but translating them into systematic, scaled organizational capability remains the constraining challenge.
2025-Q1: Institutional commitments accelerate with supply-side maturation (MIT research advocates hands-on training over bans; RAI Institute launches RAISE AI Pathways Program for operational training; IDC reports 75% of responsible AI adopters show improvements). However, demand-side paradox persists: 87% of leaders see responsible AI as essential but 85% feel unprepared, 76.5% of Asia/Pacific enterprises cannot detect AI attacks. Real-world deployment (AltaML) demonstrates maturity assessment benefits. Regulatory and litigation risks mount, intensifying training urgency. The constraint has evolved from resource allocation clarity (2024) to operational execution and expertise shortage—organisations recognise training criticality but struggle with systematic, scaled deployment.
2025-Q2: Professional certification infrastructure accelerates—ISACA launches Advanced in AI Audit (AAIA) credential in May; RAISE Pathways Program expands with five-level progression and external verification badges; university-community college partnerships broaden curriculum access. However, enterprise execution gaps widen: ISACA survey shows 89% of digital trust professionals need AI training within two years, yet only 28% of organisations train all employees (EY, June); 85% of enterprise leaders report unprepared to operationalise responsible AI (Tredence, May); only 12% have mature governance frameworks. Paradox sharpens: infrastructure expands while execution remains constrained by expertise shortage and organisational readiness gaps.
2025-Q3: Supply-side certification and training infrastructure matures—Microsoft refreshes certification landscape toward AI-centric credentials; ISACA launches AAISM (Advanced in AI Security Management); Northeastern formalizes structured responsible AI practice frameworks; Azure ML integrates RAI assessment tools; peer-reviewed research synthesizes ethical frameworks with governance models. However, demand-side execution reveals structural gaps: Infosys study (1,500 large enterprises) finds only 2% meet RAI standards, 95% experience AI incidents, 77% report financial losses; synthesis of multiple studies shows 93% use AI but only 7% have embedded governance; 62% lack documented governance plans. Paradox inverts: problem is no longer training availability but organisational readiness to embed governance at scale.
2025-Q4: Supply-side certification and training infrastructure reaches saturation with ISACA completing three-credential suite (AAIA, AAISM, AAIR), IEEE launching responsible procurement training aligned to IEEE 3119, Microsoft completing AI-centric certification refresh, and enterprise maturity assessment services operationalized (Accenture, Northeastern, RAI Institute). However, workforce readiness crisis deepens: National Academies reports 47% monthly AI use but 30% anxiety about falling behind; Bright Horizons survey finds 79% of workers unprepared and 65% received no training; Wharton study reveals training investment declining despite 74% ROI claims. MIT SMR identifies structural obstacles preventing implementation despite framework abundance. By Q4, constraint has shifted entirely from supply (resolved) to organisational culture, embedding capability, and resource allocation for systematic, scaled deployment.
2026-Jan: Microsoft launches AI Transformation Leader certification for business leaders emphasizing responsible AI governance (Jan 2026). Nasscom India survey (conducted Oct-Nov 2025, reported Jan 2026) reveals paradox: 90% of enterprises invest in RAI training but only 30% achieve maturity; skill shortages block operationalization. Market forecast predicts 75% of large enterprises will require supplier RAI certifications by end of 2026. Real-world deployment signals: Big Four bank validates GenAI through RAISE Pathways verification; Accenture's scaled mandatory training demonstrates enterprise-wide operationalization. New risk exposure: 600+ AI hallucination cases implicating legal professionals, driving legal sector training urgency and governance reinvigoration. Workforce gaps persist: 79% unprepared despite accelerating adoption. Supply infrastructure mature; execution and embedding remain the bottleneck.
2026-Feb: Supply-side infrastructure accelerates with Microsoft launching four new AI certifications including AI Business Professional and AI Transformation Leader (Feb 26); EC-Council releases CRAGE (Certified Responsible AI Governance & Ethics) credential addressing 700K U.S. reskilling gap and $5.5T global risk exposure. NIST announced listening sessions (April 2026) to identify sector-level adoption barriers. However, demand-side reality persists: Economist Impact survey shows only 4% achieve AI ROI, 16% use structured training, 8% have governance frameworks—quantifying enterprise execution and investment gaps. PwC survey provides contrasting signal: 58% report RAI improves ROI, 61% at strategic/embedded stage. MIT research cites 95% failure rate for GenAI projects with 61% of leaders under pressure to prove ROI. Pattern clear: supply accelerating while enterprise capability and measurement discipline lag; training infrastructure exists but operationalization at scale remains constrained by execution discipline and resource allocation.
2026-Apr: Certification ecosystem expanded further with EC-Council's CRAGE and CAIPM credentials, CMI's Level 7 Strategic Leadership of AI qualification, and a full Microsoft portfolio refresh of 10+ AI-specific credentials, with Coursera reporting 234% YoY growth in generative AI enrollments. The U.S. government committed $369M via DOL/NSF TechAccess to standardize AI literacy across 56 state hubs. Against this supply expansion, Stanford's 2026 AI Index documented AI incidents surging 55% year-over-year (362 in 2025 vs. 233 in 2024) and recommended certification programs as core organizational competency—while Docebo's enterprise learning report found 85% of employees say training does not help them apply AI to actual work and only 9% of organizations have training aligned to business strategy, sharpening the structural gap between credential proliferation and on-the-job capability transfer.
2026-May: Supply-side credentialing expanded further with IEEE launching CertifAIEd, a comprehensive ethical AI certification program with professional, product, and curriculum tracks. Government mandates operationalized: Texas enacted FY26-27 criteria requiring annual certified AI awareness training for all state employees; Washington D.C. launched mandatory responsible AI training policy for city government workers. Deloitte's survey of 3,235 enterprise leaders across 24 countries documented only 20% reporting high talent preparation despite 60% workforce tool access—deepening the global readiness gap. Critical negative signal intensified: only 4% of organizations deliver role-specific training and 85% of learners report training fails to help them apply AI to actual work (Docebo, 2,000+ respondents), confirming that certification proliferation has not resolved the fundamental performance transfer problem.
2026-Jun: Asia-Pacific governments accelerated mandatory training mandates—India NAIDM, Singapore's 40,000-person program, and Philippines certification standards all launched, signaling regional governance operationalization beyond the EU/US axis. Joint Commission (evaluating 23,000+ healthcare organizations) launched RUAIH certification with five governance competency requirements, validating sector-specific credentialing in regulated industries. Structured upskilling demonstrated measurable ROI: Workera's 88,000-assessment benchmark showed Responsible AI competency moving from 25% to 81% accomplished post-training, while a TVET pilot achieved 70.6% completion and 90% confidence gains. Against this, McKinsey/WRITER survey of 2,400 executives identified knowledge and training deficits as the #1 barrier to responsible AI deployment, and Docebo's enterprise learning data confirmed 85% of learners find training unhelpful—sustaining the structural gap between credential supply and on-the-job performance transfer.
2026-Jul: Vendor certification ecosystem restructured at mid-year: Microsoft retired its foundational AI/ML certifications (Azure Data Scientist, AI Engineer, AI Fundamentals, all June 2026) in favor of agentic-AI and domain-specialized credentials, signaling a sector-wide pivot away from general upskilling toward deployment-specific competency. EU AI Act Article 4 compliance urgency (August 2 deadline) drove adoption of role-based frameworks, with mature structured programs demonstrating 42% ROI versus 21% baseline (DataCamp, 500 enterprise leaders); however, the talent capacity gap deepened—IDC's 2026 Global IT Skills Survey found only one-third of IT/business leaders ready for AI integration, and independent analysis documented only 1 qualified AI engineer available per 10 open GenAI positions, confirming that credential proliferation has not resolved builder shortages. Independent verification emerged as an ecosystem priority: a 17-author peer-reviewed analysis argued responsible AI practices produce no external market signal absent outcome-oriented, independently-verified certification, while the OECD's 38-country assessment (68,138 data points) found AI adoption jumped from 7% to 20% of firms since 2021 but only ~1% of the workforce holds advanced capabilities. Governance-specific skill shortages sharpened: an analysis of 3,519 EU job postings found 6.7 builder roles for every governance role (ranging 3.5:1 to 16:1 by country), with 71.5% of governance postings omitting any EU AI Act reference despite August enforcement, and Innovate UK's sector white paper (65M job postings analyzed) flagged responsible-AI and governance training as a priority gap. The European Commission committed to AI-aware workforce development via its Cybersecurity Skills Academy, with training modules due Q4 2026, reinforcing policy-level investment alongside the persistent execution gap.
2026-Aug: Regulatory compliance deadline (EU AI Act Article 4, August 2) drove market urgency for documented, role-specific AI literacy training; 79% of enterprise job postings now cite AI literacy as requirement (VDF AI analysis), signaling broad compliance demand. Vendor supply-side acceleration continued: Microsoft launched Enterprise AI Governance, Ethics & Security Professional Certificate on Coursera (July 2026, 5-course program with NIST/ISO/GDPR coverage); EC-Council scaled CRAGE to North America with 11-module governance curriculum addressing ISO 42001 standards. However, independent research reinforced effectiveness gap: Conference Board study (1,300 employees, 35 leaders) found 55% use AI weekly/daily but only 33% received formal training with time/resources constraints; Schellman's ANAB-accredited governance maturity survey (525 US respondents) revealed confidence-maturity gap (90% funded governance programs, 74% claim audit-readiness, only 27% actually mature), though mature governance correlated with 78% production AI agent deployment. Critical synthesis analysis (Mia AI, drawing from DataCamp/BCG/MIT/IDC) identified root cause as structural learning gap: 82% of organizations provide training yet 59% report skills gaps, with only 5% achieving substantial financial gains. UK government guidance (DWP/Skills England) validated effective approaches as embedded, task-based, structured training linked to actual work—identifying system-level training infrastructure as too slow to match adoption pace. Paradox persists at scale: training infrastructure matures while organizational execution discipline, role-differentiation, and behavior-change measurement remain constrained.