> ## Content Index
> Fetch the complete content index at: https://signal-engine.ghost.io/llms.txt
> Use this file to discover other available public pages before exploring further.

# The AI training gap: what businesses say is hardest right now, and how to teach the shop floor
- URL: https://signal-engine.ghost.io/ai-training-gap-shop-floor-to-management/
- Published: 2026-11-15T14:04:35.000Z
- Updated: 2026-11-15T14:04:35.000Z
- Description: Surveys from 2025-2026 show AI stalling on skills, failed pilots, weak ROI, shadow tools and a floor-to-office gap. Here's how small shops can find trainers and teach the link that matters.
- Author: John Caraher
- Tags: Featured, Issue 2, Trending Now, AI, Small Business, #Import 2026-11-15 14:13

Updated 15 Nov 2026, 15:59 SAST

Contents

1. [The short version](#short-version)
2. [What businesses say is hardest right now](#hardest-right-now)
3. [The skills gap, and the shortage of people who can teach](#skills-gap)
4. [Why training matters most between the shop floor and management](#shop-floor-to-management)
5. [How to find and vet a good trainer](#finding-trainers)
6. [What the South African numbers show](#south-africa)
7. [What to watch next](#what-to-watch)
8. [Why it matters for your business](#why-it-matters)
9. [Sources](#sources)

**The short version:** Companies are buying AI tools faster than they can get them to work. Surveys from McKinsey, Gartner, BCG, the OECD, Microsoft and others, dated from 2025 into late 2026, point to the same problems: skills shortages, a lack of people who can train teams, pilots that never leave the trial stage, weak returns, messy data, staff using unapproved tools, and a gap between leaders and the floor. Training is the part most firms underdo. In a small or mid-sized shop, the training that matters is not a general “intro to ChatGPT” session. It is teaching operators, supervisors and managers how to use — and check — the same tools on the same jobs.

## What businesses say is hardest right now

The complaints below are from named, dated surveys and reports. They are not a single global league table. Each figure belongs to its own sample, and the samples differ in size, country and the kind of company asked. Where two reports give different numbers for a similar idea, both are given, not averaged.

### Pilots that never scale

McKinsey’s [State of AI global survey for 2025](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-2025?ref=signal-engine.ghost.io) found that 88 percent of respondents said their organisations regularly use AI in at least one business function, up from 78 percent a year earlier. Most were still experimenting or piloting. McKinsey wrote that “approximately one-third” had begun to scale, and that “nearly two-thirds of respondents say their organizations have not yet begun scaling AI across the enterprise.” Sixty-two percent were at least experimenting with AI agents. Sixty-four percent said AI was enabling innovation, but only 39 percent reported an EBIT impact at the enterprise level.[\[1\]](#fn-1) A 2 March 2026 McKinsey note said “no more than one in 10 report AI agent usage making it past the pilot stage within a specific business function.”[\[2\]](#fn-2) A separate McKinsey piece on data readiness said that bottleneck is “a key reason only 7 percent of companies have fully scaled AI across their organizations.” That 7 percent figure is not the same as the “begun to scale” finding, and the two should not be treated as one number.[\[3\]](#fn-3)

Gartner, in a 12 June 2025 note based on its 2024 AI Mandates for the Enterprise Survey, said that on average “only 41% of generative AI prototypes and 42% of nongenerative AI prototypes reached production.”[\[4\]](#fn-4) A 9 May 2025 Gartner note, aimed at technology-services firms, said: “Sixty percent of GenAI pilots fail to enter production.” That is a finding in that client context, not a restatement of the 41/42 percent figures.[\[5\]](#fn-5) In a January–April 2026 survey of 1,303 respondents at organisations with at least $50 million in 2025 revenue, only 22 percent “have successfully scaled AI across multiple business units or adopted an AI-first approach.” Roughly 11 percent were “entirely unaware of what their function spent on AI in 2025.” Eighty-five percent of functional leaders planned to increase spending in 2026, after dedicating an average of 12 percent of functional budgets to AI in 2025.[\[6\]](#fn-6)

### Cost, value and ROI

On 19 October 2026 Gartner reported a survey of 2,002 CIOs and technology executives, fielded May to July 2026\. Only 13 percent of CIOs said their enterprise had “realized significant or transformational value from AI so far.” The likelihood of use-case success “remains one-in-five.” Fifty-eight percent of CIOs faced increased pressure to deliver cost improvement from AI, while 51 percent expected AI to increase the total cost of ownership of technology solutions. Only 31 percent had guidelines for when an AI pilot should not move to production, and only 17 percent had guidelines for abandoning low-value AI already in production. Gartner said organisations with the first kind of guideline were 2.3 times more likely to be what it calls “vanguards,” and those with the second were 2.6 times more likely.[\[7\]](#fn-7) In the earlier 2026 functional-leader survey, Gartner said “high performers” who track ROI and stop weak projects reported positive returns in 81 percent of AI initiatives, while “low performers” did not know the return for 29 percent. Productivity was a key target for 75 percent of functional leaders.[\[6\]](#fn-6) That sits alongside McKinsey’s 39 percent enterprise-level EBIT figure and the South African ROI figures [below](#south-africa).

### Skills, training time and the trainer shortage

In BCG’s global survey of 1,400 C-suite executives, 62 percent “cited a shortage of talent and AI skills as their biggest challenge to achieving AI value,” BCG wrote in 2025 — ahead of unclear priorities or lack of strategy. In the same piece, BCG said only 6 percent of those executives “have begun upskilling their workforce in a meaningful way,” and that respondents said “less than 25% of their learning time occurs during work hours.”[\[8\]](#fn-8)

McKinsey’s January 2025 workplace report, *Superagency in the workplace*, said nearly half of employees wanted more formal training and believed it was the best way to boost AI adoption, while “more than a fifth report that they have received minimal to no support.” It also said 46 percent of leaders identified skill gaps as a significant barrier to AI adoption.[\[9\]](#fn-9) A McKinsey “AI upskilling challenge” note said 75 percent of US workers anticipated their roles changing because of AI in the coming five years, but only 45 percent had gone through a recent upskilling programme.[\[11\]](#fn-11)

The World Economic Forum’s [Future of Jobs Report 2025](https://reports.weforum.org/docs/WEF%5FFuture%5Fof%5FJobs%5FReport%5F2025.pdf?ref=signal-engine.ghost.io) said that if the world’s workforce were 100 people, 59 would need training by 2030; employers foresaw 29 being upskilled in their current roles and 19 reskilled and redeployed, while 11 “would be unlikely to receive the reskilling or upskilling needed.” Employers expected 39 percent of workers’ core skills to change by 2030\. In the Forum’s Executive Opinion Survey, “half of executives worldwide highlight a lack of skills to support adoption as the top barrier,” followed by “a lack of vision among managers and leaders (43 percent).” Seventy-seven percent of surveyed employers planned, by 2030, to reskill or upskill people to work alongside AI.[\[12\]](#fn-12)

The OECD, in its 2025–2026 work on AI, skills and SMEs, said around 40 percent of employers in manufacturing and finance who had not yet adopted AI named skills as the main reason, “as do more than half of SMEs that are not yet using generative AI.” Fewer than 1 percent of workers will need advanced AI-specific skills such as programming or model development; most need digital skills, the ability to interpret data, and human skills such as problem-solving.[\[13\]](#fn-13) Among SMEs that already use generative AI, only 23.6 percent reported that employees take part in AI-related training (from 11.3 percent in Japan to 29.4 percent in Canada). For firms not using it, that share fell to 2.7 percent, “despite lack of skills being a commonly reported barrier.”[\[14\]](#fn-14)

### Resistance, fear and change management

BCG has argued that the hardest adoption problems are often human, not technical. In a 2025 article, it said “cultural resistance and emotional friction—not tech—are the biggest blockers to adoption,” and quoted its researchers on “psychological barriers—things like fear, habit, or just knowing what’s possible with the technology,” including worries about job security.[\[17\]](#fn-17) In its “AI adoption puzzle” analysis, BCG said the most frequent issues clustered around “a lack of trust in AI and limited time to upskill, followed by cybersecurity and copyright concerns, skill atrophy and job security fears, and a lack of training resources.”[\[8\]](#fn-8) In a December 2025 software-delivery study, BCG said only about 40 to 50 percent of organisations reported “substantial or moderate investment” in change management, and about 27 percent cited IT or cyber approval as the most time-consuming bottleneck. Those figures come from a software-delivery sample, not a survey of factory floors.[\[18\]](#fn-18)

Microsoft’s 2026 Work Trend Index said organisational factors — “culture, manager support and talent practices” — “account for more than 2× the reported AI impact of individual factors like mindset and behaviour (67% vs. 32%).” Among AI users, 50 percent ranked “quality control of AI output” and 46 percent ranked critical thinking as the human skills that matter more as AI takes on more work; 86 percent said they treat AI output as a starting point and “stay responsible for the thinking.” Only 19 percent of AI users sat in what Microsoft called the “Frontier” group, where individual skill and organisational support are both high.[\[19\]](#fn-19)

### Shadow AI, data and the leadership–frontline split

“Shadow AI” means staff using AI tools that the company has not approved or does not even know about. An IBM-sponsored Censuswide study, described by IBM on 3 November 2025, surveyed 3,000 full-time office workers in North America who were familiar with AI tools; the US figures are based on 1,000 American respondents. IBM said 80 percent of American office workers in that sample used AI in their roles, but “only 22% rely exclusively on tools provided by their employers.” Among Gen Z respondents aged 18–24, 35 percent said they were likely to use only personal AI applications, compared with 14 percent in other age groups.[\[20\]](#fn-20)

Cisco’s 2025 Cybersecurity Readiness Index, based on a survey of 8,000 private-sector security and business leaders in 30 markets, said 51 percent of employees use third-party generative AI tools and 22 percent have unrestricted access to public tools, while “60 percent of IT teams are unaware of how employees interact with generative AI tools” and 60 percent of organisations “lack confidence in their ability to detect unregulated or shadow AI deployments.”[\[22\]](#fn-22) PagerDuty, on 11 June 2026, published a Wakefield Research survey of 1,250 office professionals in non-IT roles at organisations with at least $500 million in revenue in the United States, the United Kingdom, Australia and Japan. It said 66 percent had used AI tools at work even though they believed that was not permitted, and 81 percent believed “leadership operates under a different set of rules than the rest of the company when it comes to AI.” Seventy-five percent said they would be likely to look for a new job that offered better AI skills development. That sample is large-company office staff, not small manufacturers.[\[23\]](#fn-23)

McKinsey has also described a readiness gap in the other direction: 70 percent of respondents felt “personally prepared to adopt and use AI,” but “only 27 percent of leaders believe their organizations are ready to make the shifts needed for an agentic future.” McKinsey’s reading was that “employees are adapting to AI faster than the institutions they work in.”[\[24\]](#fn-24) On data, McKinsey tied a “governed, traceable, and reusable foundation” to the 7 percent “fully scaled” figure already noted. The OECD said more than half of SMEs (54 percent) reported concerns about copyright, legal or regulatory issues “and about what happens to the information fed into generative AI models.” Quality of output was a concern for 35 percent.[\[14\]](#fn-14)

## The skills gap, and the shortage of people who can teach

Put those surveys side by side and a pattern appears. Firms say they cannot find enough people with AI skills. Employees say they want formal training and are not getting much of it. Executives, in BCG’s figures, rarely say they have started meaningful upskilling. SMEs, in the OECD’s figures, rarely train even when they already use the tools. That is the trainer shortage in practice: not only a lack of machine-learning engineers, but a lack of people who can sit with a working team, on real work, and teach them how to use a tool without breaking a process or leaking a customer file.

The OECD’s point that fewer than 1 percent of workers need advanced AI-coding skills matters for a press shop. You are not, in the main, hiring a model-builder. You need people who can judge an output, keep a clean data trail, and explain a change to the next shift. Microsoft’s 2026 index, already noted, ranks quality control of AI output and critical thinking as the human skills users themselves value most. Training that only teaches people to “ask the chatbot better” is aiming at the wrong target, if those surveys are right.

## Why training matters most between the shop floor and management

McKinsey’s work on US frontline workers is one of the few recent reports that talks directly about the people who make, move or sell a product. It said “almost half of US C-suite executives report that their deployment of AI tools is too slow, and the number one reason they offer for the delay is ‘talent skill gaps.’” Nearly all employees were experimenting with generative AI tools, McKinsey said, “but since only a small fraction of organizations have built structured talent strategies to improve AI fluency, it’s hard for workers to do more with the technology.”[\[25\]](#fn-25)

In manufacturing specifically, McKinsey said “nearly three-quarters of workers we surveyed report experiencing skill gaps,” both in foundational operational skills such as equipment maintenance and repair, and in core craft or technician skills. Leaders and employees disagreed about what to teach: 65 percent of surveyed leaders listed technological skills as one of the two most important types to invest in, but only 35 percent of employees agreed. Fifty percent of employees listed socioemotional skills among their top two, compared with 36 percent of leaders. Manufacturing COOs, McKinsey said, listed “cultural shift” and “reskilling” among the top three barriers to implementing AI. McKinsey, citing a recent study, also wrote that companies spend an estimated $9,100 a year on software-as-a-service per employee, and just $1,200 per employee on training. That comparison is McKinsey citing another study, not a figure it surveyed in that article.[\[25\]](#fn-25)

Training that only happens in the office makes this worse. Managers learn a demo; operators hear a rumour. Shadow AI then fills the hole, as the IBM, Cisco and PagerDuty figures show in office samples: people use whatever tool they already know, including on the wrong data. The useful design, for a shop, is a single training plan with three rooms, or three hours, aimed at different jobs:

- **Operators and frontline staff** need practice on the actual job in front of them — a job card, a breakdown, a quality check — and a simple rule for what may and may not be entered into a tool. They also need permission to say when the output is wrong.
- **Supervisors** need to learn the handoff: when to accept a suggestion, when to override it, and how to brief the next shift in plain language.
- **Owners and managers** need to learn enough to set the standard, pick the tool, pay for practice time during work hours, and stop a pilot that is not earning its keep — the “stopping rules” Gartner says most CIOs still do not have.

## How to find and vet a good trainer

Analysis The surveys do not offer a directory of trainers. They do show what weak training looks like: one-off classroom events, learning pushed into personal time, and generic demos that never touch a real workflow. The checks below are practical, not a certified standard.

- **Have they taught a shop like yours?** Ask for a previous client in a similar trade — a toolroom, a press shop, a fabricator, a small manufacturer — and call that person. A trainer who has only taught office teams how to write emails is not yet proven on the floor.
- **Will they work on your jobs, not a slide deck?** The test is whether they will sit with a real quote, job card, downtime log or quality sheet. If the session cannot use your documents (with sensitive fields removed), it is a product demo.
- **Can they teach operators and managers in the same programme?** Not the same lecture, but linked sessions, so both sides leave with the same rules about data, checking and when to override.
- **Do they leave you with a short written rule-book?** What may be pasted into a tool, who checks the output, what is recorded, and what happens when the tool is wrong. If they cannot put that on one or two pages, the training will fade at the end of the week.
- **Is practice time during paid hours?** BCG’s executives said most learning time was not during work hours. If a trainer only offers after-hours webinars, expect weak uptake on the floor.
- **Are they selling a licence, or a skill?** A vendor can still be a decent teacher, but you should know which hat they are wearing. Be wary of guaranteed savings, claims that “AI will replace the team,” or a refusal to talk about failure cases.

Good trainers are scarce because the skill is scarce, which is what the C-suite and SME surveys are already saying. If you cannot find one immediately, a stopgap is to pick one process, appoint an internal “checker” on the floor and one in the office, and have them learn the same tool together for a month, then teach the next two people. That is slower than a polished course. It is closer to what the high-performing firms in these reports describe: practice, stopping rules, and managers who use the tool themselves.

## What the South African numbers show

SAP Africa, reporting in May 2025 on research done at the end of 2024 among mid-size and enterprise companies in Kenya, Nigeria and South Africa, said every organisation surveyed expected demand for AI skills to rise in 2025 and expected some AI-related skills gap. Organisations in South Africa were “most likely to experience AI skills shortages (53%), followed by Nigeria (50%) and Kenya (43%).” Eighty-five percent named AI development skills as a priority and 83 percent named generative-AI skills. Where skills were simply not available, 68 percent of South African respondents were using automation to augment their AI skills, SAP said, the highest of the three countries.[\[26\]](#fn-26)

PwC’s South African work on AI in manufacturing operations said only 13 percent of local respondents — and 4 percent globally — reported that they were already achieving significant financial benefits and ROI, “underscoring a relatively low AI maturity level.” About 25 percent of South African responses qualified as “AI Operations Champions,” against a global 8 percent in that study. PwC said that locally, “the cost of AI software and the lack of specialised AI experts are the primary concerns,” along with a lack of AI knowledge and resistance to change.[\[27\]](#fn-27)

World Wide Worx’s *South African Generative AI Roadmap 2025* said the most common skills approach, chosen by 53 percent of respondents, was upskilling current employees; 34 percent used internal training programmes; 22 percent were hiring AI specialists; 22 percent were collaborating with educational institutions; and 22 percent had “no specific measures in place.” Among companies already using generative AI, 37 percent reported having no safeguards, and only 13 percent had comprehensive guardrails addressing safety, privacy and bias.[\[28\]](#fn-28)

Closer to the factory gate, a 2026 skills audit by the Nelson Mandela Bay Business Chamber and Harambee, covering 55 manufacturing enterprises and 12,662 employees, reported that 80.4 percent of companies surveyed said their existing workforce “lacks the foundational digital skills required to transition into new technology-enabled roles.” The audit set a Hard-to-Fill Index baseline of 30.4, classified as critical pressure. The future skills it listed were practical — PLC and automation diagnostics, electro-mechanical multi-skilling, predictive maintenance, digital quality and IT/OT integration — not prompt-writing.[\[29\]](#fn-29) That study is about one metro’s manufacturers, and it covers digital and automation skills as well as AI. It is still a clear local signal that the floor is short of the basics a modern tool needs.

## What to watch next

Watch the next McKinsey State of AI and Gartner CIO surveys, to see whether the share of firms that have actually scaled — and the 13 percent who report significant value — moves, and whether OECD SME training rates rise above the low 20s among users. In South Africa, watch whether manufacturer skills audits, like the Nelson Mandela Bay barometer, are repeated. We will update this piece if a major new survey changes the picture.

## Why it matters for your business

Analysis If you run a small or mid-sized shop, these surveys were mostly not written about you. Many of the samples are large companies, office professionals or global executives. The pattern still travels. Buying a tool is the easy part. Getting a night-shift operator, a supervisor and an owner to use it on the same job, with the same checks, is the hard part. That is what “pilots that never scale” looks like in a factory of 40 people.

The practical response is not to hire a data-science team you cannot staff, or to wait for the perfect trainer. It is to treat training as part of the purchase price: paid time on the floor, a written rule about data, a named person who checks outputs, and a date on which you will stop the pilot if it has not earned its keep. Train the link between the floor and the office first. The global numbers say most organisations still fail at that step. Doing it on one process — quoting, scheduling, maintenance or quality — is not a claim that AI will raise your profits. It is a reason to spend as carefully on teaching as on software.

This article is for information only and is not financial or investment advice.

Spotted an error? Tell us and we'll correct it.

### Sources

1. [McKinsey, The State of AI: Global Survey 2025](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-2025?ref=signal-engine.ghost.io) [↩](#fnref-1)
2. [McKinsey, 2 March 2026: “Are your people ready for AI at scale?”](https://www.mckinsey.com/capabilities/people-and-organization/our-insights/the-organization-blog/are-your-people-ready-for-ai-at-scale?ref=signal-engine.ghost.io) [↩](#fnref-2)
3. [McKinsey: “AI data readiness: Foundation for scaling enterprise AI”](https://www.mckinsey.com/capabilities/mckinsey-technology/our-insights/ai-data-readiness-the-key-to-scaling-impact?ref=signal-engine.ghost.io) [↩](#fnref-3)
4. [Gartner, 12 June 2025: “AI Maturity Matters: Proportion of AI and GenAI Prototypes Making It Into Production”](https://www.gartner.com/en/documents/6587902?ref=signal-engine.ghost.io) [↩](#fnref-4)
5. [Gartner, 9 May 2025: “Align Your Strategy to Client Maturity to Scale Generative AI Engagements”](https://www.gartner.com/en/documents/6456307?ref=signal-engine.ghost.io) [↩](#fnref-5)
6. [Gartner newsroom: January–April 2026 survey, “Only 22% of Organizations Have Successfully Scaled AI Across Multiple Business Units”](https://www.gartner.com/en/newsroom/press-releases/gartner-survey-finds-only-22-percent-of-organizations-have-successfully-scaled-ai-across-multiple-business-units?ref=signal-engine.ghost.io) [↩](#fnref-6)
7. [Gartner newsroom, 19 October 2026: “Only 13% of CIOs Say Their Enterprise Has Realized Significant or Transformational Value from AI”](https://www.gartner.com/en/newsroom/press-releases/2026-10-19-gartner-survey-finds-only-13-percent-of-cios-say-their-enterprise-has-realized-significant-or-transformational-value-from-ai?ref=signal-engine.ghost.io) [↩](#fnref-7)
8. [BCG, 2025: “AI Adoption Puzzle: Why Usage Is Up But Impact Is Not”](https://www.bcg.com/publications/2025/ai-adoption-puzzle-why-usage-up-impact-not?ref=signal-engine.ghost.io) [↩](#fnref-8)
9. [McKinsey, 28 January 2025: “Superagency in the workplace”](https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work?ref=signal-engine.ghost.io) [↩](#fnref-9)
10. [McKinsey: “Why digital upskilling is the future of work”](https://www.mckinsey.com/capabilities/people-and-organization/our-insights/we-are-all-techies-now-digital-skill-building-for-the-future?ref=signal-engine.ghost.io) [↩](#fnref-10)
11. [McKinsey: “The AI upskilling challenge”](https://www.mckinsey.com/featured-insights/people-in-progress/the-ai-upskilling-challenge?ref=signal-engine.ghost.io) [↩](#fnref-11)
12. [World Economic Forum, Future of Jobs Report 2025](https://reports.weforum.org/docs/WEF%5FFuture%5Fof%5FJobs%5FReport%5F2025.pdf?ref=signal-engine.ghost.io) [↩](#fnref-12)
13. [OECD: “AI and skills”](https://www.oecd.org/en/publications/ai-and-skills%5Ff843b352-en/full-report.html?ref=signal-engine.ghost.io) [↩](#fnref-13)
14. [OECD, 2025: “Generative AI and the SME Workforce”](https://www.oecd.org/en/publications/generative-ai-and-the-sme-workforce%5F2d08b99d-en/full-report/component-6.html?ref=signal-engine.ghost.io) [↩](#fnref-14)
15. [LinkedIn Learning, 2025 Workplace Learning Report](https://learning.linkedin.com/resources/workplace-learning-report?ref=signal-engine.ghost.io) [↩](#fnref-15)
16. [LinkedIn, 2026 Talent Velocity Advantage Report](https://learning.linkedin.com/resources/talent-velocity-report?ref=signal-engine.ghost.io) [↩](#fnref-16)
17. [BCG, 2025: “When Companies Struggle to Adopt AI, CEOs Must Step Up”](https://www.bcg.com/publications/2025/when-companies-struggle-to-adopt-ai-ceos-must-step-up?ref=signal-engine.ghost.io) [↩](#fnref-17)
18. [BCG, December 2025: “GenAI in SDLC: What top performers do differently”](https://insights.bcg.com/rs/799-IOB-883/images/20251218%5FBCG%5FGenAIInSDLC.pdf?ref=signal-engine.ghost.io) [↩](#fnref-18)
19. [Microsoft, 2026 Work Trend Index Annual Report](https://news.microsoft.com/annual-work-trend-index-2026/?ref=signal-engine.ghost.io) [↩](#fnref-19)
20. [IBM, 3 November 2025: “Is rising AI adoption creating shadow AI risks?”](https://www.ibm.com/think/insights/rising-ai-adoption-creating-shadow-risks?ref=signal-engine.ghost.io) [↩](#fnref-20)
21. [IBM Canada, 3 September 2025: Shadow AI study (Censuswide, May 2025 fieldwork)](https://canada.newsroom.ibm.com/2025-09-03-IBM-R-Study-Shadow-AI-Use-Surges-as-Canadian-Workers-Outpace-Employers-in-AI-Adoption?ref=signal-engine.ghost.io) [↩](#fnref-21)
22. [Cisco 2025 Cybersecurity Readiness Index, as reported in coverage of the survey of 8,000 leaders in 30 markets](https://petri.com/cisco-businesses-unprepared-ai-cyberattacks/?ref=signal-engine.ghost.io) [↩](#fnref-22)
23. [PagerDuty, 11 June 2026: Shadow AI Workplace Survey (Wakefield Research)](https://www.pagerduty.com/newsroom/shadow-ai-workplace-survey-2026/?ref=signal-engine.ghost.io) [↩](#fnref-23)
24. [McKinsey: “Closing the AI readiness gap”](https://www.mckinsey.com/capabilities/people-and-organization/our-insights/from-adoption-to-impact-three-horizons-of-ai-transformation?ref=signal-engine.ghost.io) [↩](#fnref-24)
25. [McKinsey: “Investing in frontline workers’ AI skills”](https://www.mckinsey.com/capabilities/operations/our-insights/a-us-productivity-unlock-investing-in-frontline-workers-ai-skills?ref=signal-engine.ghost.io) [↩](#fnref-25)
26. [SAP Africa, May 2025: “Does Africa Have the Skills to Benefit from AI?”](https://news.sap.com/africa/2025/05/does-africa-have-the-skills-to-benefit-from-ai/?ref=signal-engine.ghost.io) [↩](#fnref-26)
27. [PwC South Africa: “AI in operations: Revolutionising the manufacturing industry”](https://www.pwc.co.za/en/publications/ai-in-operations.html?ref=signal-engine.ghost.io) [↩](#fnref-27)
28. [World Wide Worx, The South African Generative AI Roadmap 2025](https://www.worldwideworx.com/wp-content/uploads/2025/10/The-SA-Gen-AI-Roadmap-2025.pdf?ref=signal-engine.ghost.io) [↩](#fnref-28)
29. [Nelson Mandela Bay Business Chamber and Harambee, 2026 manufacturing skills audit](https://www.nmbbusinesschamber.co.za/news/bay-s-manufacturing-future-hinges-on-closing-the-skills-gap?ref=signal-engine.ghost.io) [↩](#fnref-29)