Inclusive artificial intelligence for industry

UNCTAD does more than warn it offers a roadmap. Here’s how industries, particularly in emerging economies and among private partners, can translate these recommendations into action. 

Fair access to technology and infrastructure 

Invest jointly in regional infrastructures (shared computing centers, data hubs) instead of reproducing centralized architectures. 

Promote “Edge + Cloud” models that reduce dependence on costly international connections. 

These measures lower entry barriers and broaden the number of actors capable of industrializing AI locally. 

Training, skills and capacity-building strategies 

The true engine of inclusive AI is neither algorithms nor data centers but human skills: those who design, adapt and operate them. 

AI only holds value insofar as it integrates into professions, decisions and the daily gestures of industry. 

That’s why UNCTAD insists: without large-scale skill development, AI risks becoming a factor of exclusion rather than progress. 

But what does “capacity-building” actually mean in practice? 

1.Train differently, train together 

Companies can no longer rely on top-down or generic training. What’s needed are targeted, operationally grounded programs: upskilling technicians, industrial bootcamps for engineers, hands-on workshops for managers. 

These initiatives have the greatest impact when co-created with economic actors manufacturers, technical schools, professional associations. 

2.Learning by doing and building

Here comes the well-known principle of “learning by doing”: experiment, test, iterate. 

Put employees at the center of pilot projects, allow them to handle data, explore real use cases of automation or predictive analysis. 

Sectoral incubators — laboratories where engineers, data scientists, and business leaders meet — play a key role. They turn curiosity into competence, and competence into innovation. 

In truth, the best way to learn AI is to make it speak to your profession. 

• How can a line operator use a predictive interface? 
• How can a logistician interpret a model’s recommendations? 
• How can a plant manager balance automation with the preservation of know-how? 

Such questions find answers only through practice not manuals. 

3.Rethinking Organization and HR Pathways 

The challenge is also organizational. 

Integrating AI into existing structures requires rethinking HR pathways, evaluation systems, and career plans. 

An “AI-ready” company is not one that hires data scientists by the dozen, but one that develops hybrid skills at the intersection of domain expertise and data literacy. 

We’re talking about training internal translators: profiles able to understand on-the-ground constraints while communicating effectively with technical experts. These intermediaries become essential conduits for intelligence. 

4.Building a Culture of Continuous Learning 

Finally, AI evolves too quickly for training to remain a one-off event. 

A culture of continuous learning must be built one that encourages curiosity, views experimentation not as a risk but as a necessity. 

Leaders play a decisive role here: they must embody openness, support internal initiatives, and value atypical career paths. 

Because ultimately, an inclusive AI is not measured by the number of deployed models, but by the quality of the humans who accompany them. 
And that quality must be cultivated, patiently, collectively.