Ai Skills: Companies Want Them, Locations Try To Provide Them, But What Exactly Are They?

As companies scramble to present AI into everything from their workflows to back-end operations, upskilling programmes will apt play a domiciled successful driving investment.

UK Prime Minister Keir Starmer delivers a reside astatine 10 Downing Street during a reception astatine London Tech Week connected 10 June 2025. Credit: JORDAN PETTITT/AFP/Getty Images.

On nan poster-sized bingo card of AI buzzwords – including, but not constricted to, agentic AI, antiauthoritarian AI, augmentation, automation and superintelligence – location is 1 that is simply a apical interest for some companies and governments astir nan world: AI skills.

More than three-quarters of companies look an AI skills shortage, according to US IT services institution UST. The skills requirements for AI-exposed sectors are changing 66% faster than for different sectors, according to a PwC study.  

Discover B2B Marketing That Performs

Combine business intelligence and editorial excellence to scope engaged professionals crossed 36 starring media platforms.

Find retired much

To plug this gap, countries person begun launching nationalist upskilling programmes to amended accommodate their respective workforces to nan needs of a changing labour market. These programmes are apt to summation a heightened relevance successful overseas nonstop investment, arsenic companies commencement to facet successful nan expertise of a section workforce to successfully usage AI devices to thrust business outcomes. In nan coming years, nan meaning of a skilled workforce will apt evolve, incorporating nan expertise to leverage AI.

However, arsenic everyone clamours to study AI skills, a important mobility is often near retired of nan debate: what precisely are they? Is location a benchmark that we tin measurement people’s skills against? When tin a country, aliases a worker, declare to beryllium AI literate?  

Answering these questions is important to nan designers and beneficiaries of AI upskilling programmes. If circumstantial outcomes aren’t outlined, governments consequence wasting money connected programmes pinch undefined intends and missing a important opportunity to amended their workforce and pull investment.  

What is AI upskilling?   

This past summer, nan UK Government announced a business pinch awesome exertion companies specified arsenic Amazon, BT, Google, IBM, Microsoft and Sage, “to train 7.5 cardinal UK workers successful basal AI skills”. This follows connected from nan government’s AI Opportunities Action Plan, launched successful January, which it presented arsenic a roadworthy representation for nan UK to go a world leader successful AI.  

The Department of Business and Trade claims each of nan participating companies will “bring a different area of expertise to its work”, pinch Microsoft already committing to upskilling 1 cardinal workers by nan extremity of 2025. Another initiative, a £187m ($253m) programme called TechFirst, intends to bring AI skills training to “classrooms and communities and train up group of each ages and backgrounds for nan tech careers of nan future”.  

Glynn Townsend is nan elder head of acquisition services astatine SAS, 1 of nan providers partnering pinch nan authorities to present this training to 7.5 cardinal workers. While he emphasises that AI literacy is simply a difficult word to define, fixed that astir group brushwood AI done nan usage of large-language models (LLMs), a person’s expertise to understand nan imaginable benefits and limitations of chatbots could beryllium 1 cardinal facet of nan term.  

“That is knowing nan bias of your models, wherever nan information [that nan exemplary has been trained on] comes from and being capable to interrogate it to make judge location is simply a statement of accuracy done it,” he outlines. “Going forward, it will beryllium astir trust, assurance and giving group nan rules pinch which they tin research to either automate aliases augment their regular activity without it being this large scary thing.”  

However, this is mostly nan lawsuit astatine nan user level. “AI literacy from a user is very different from AI literacy from personification who is going to beryllium building models comparative to what they are doing connected a regular basis,” Townsend adds.  

Rob Woodstock, managing head astatine exertion consultancy Slalom, says that knowing a worker’s domiciled wrong a company’s wider AI take strategy is important to nan type of training they should receive. Three wide categories he outlines are C-suite executives, who request an “understanding of nan imaginable of AI to heighten [their] business”; mean institution workers who tin usage AI erstwhile “driving an result aliases delivering a service”; and nan technologists and AI makers, who request to person “a really precocious level of knowing of really LLMs activity and really to build nan infrastructure that allows group to present alteration utilizing AI”.   

At nan aforesaid time, some Woodstock and Townsend item that fixed nan accelerated gait astatine which AI is evolving and being integrated into businesses, learning AI skills will person to beryllium an ongoing process. The emergence of agentic AI, for example, mightiness upend what AI literacy intends successful nan adjacent fewer years.  

“The large taste displacement has sewage to beryllium astir continual lifelong learning,” Townsend underlines. “This isn’t a one-off activity, wherever you tin present upskilling training connected AI once, because nan velocity astatine which nan exertion is moving, it is going to beryllium astir continual learning each year, each nan time.”   

The jobs question 

One obstacle to deploying skills training is people’s level of spot successful AI systems. Woodstock and Townsend opportunity reactions tin beryllium mixed. Some reason utilizing AI connected ethical grounds (Townsend said a customer erstwhile expressed doubts astir utilizing AI owed to its detrimental biology impact). Workers whitethorn besides beryllium reluctant to adopt these tools, peculiarly fixed nan definitive warnings from tech CEOs that they will frighten jobs. According to a canvass by nan Trades Union Congress, 51% of UK adults are worried astir nan effect that AI will person connected their jobs.  

These fears are not unsubstantiated, arsenic a caller study into nan effects of AI connected nan labour unit suggests that younger occupation seekers are already facing a tougher labour marketplace because of AI. Economists astatine Stanford University recovered that entry-level jobs successful nan astir AI-exposed sectors knowledgeable a 16% simplification successful nan US betwixt precocious 2022, erstwhile ChatGPT was first launched, and mid-2025. More knowledgeable workers successful these aforesaid industries, connected nan different hand, are experiencing much opportunities.   

Multiple experts interviewed by Just Food sister tract Investment Monitor reason that nan economical and governmental upheaval resulting from AI is comparable to nan disruptions caused by exertion successful nan precocious 20th century. The preamble of computers, Townsend highlights, made agency typing pools obsolete, but yet group recovered caller jobs, and nan marketplace adjusted. 

Indeed, nan World Economic Forum published a study successful January estimating that, arsenic AI transforms nan world workforce, 170 cardinal caller jobs will beryllium created by 2030. At nan aforesaid time, 92 cardinal jobs will beryllium destroyed, implying a 7% full summation successful worldwide employment of 78 cardinal jobs.

“I deliberation location is simply a very short-term disruption arsenic we are unsure of what nan effect is going to be, but I deliberation that will resoluteness itself very quickly,” Townsend says.  

Fabien Braeseman, a interrogator connected AI and activity astatine nan Oxford Internet Institute, besides interprets nan study arsenic nan short-term effects of a caller technology, and that nan occupation marketplace will yet adjust.  

“I americium reasoning this is simply a short-term observation, that group who are still training to spell into these [AI-exposed] jobs are now seeing a modulation of technological request and requirements,” he says. “AI will go arsenic modular a instrumentality arsenic smartphones are these days.” 

For Braeseman, different information successful really AI is affecting nan labour unit concerns long-standing demographic changes. In England and Wales, nan fertility rate is astatine an all-time low, meaning that successful nan agelong term, location will beryllium less group of moving age, while a greater portion of nan organization retires.  

“Demographic alteration is simply a slow process that has already reached a captious authorities successful definite occupations successful definite locations because location is conscionable not that galore group for each these jobs,” he outlines. “Maybe AI could thief america to go much effective truthful everything would much aliases little enactment unchanged.” 

However, if companies really automate astir entry-level jobs successful nan adjacent fewer years, arsenic Anthropic CEO Dario Amodei has warned they will, this could besides create a vacuum for young workers. If caller opportunities and capable training do not look quickly enough, past a bigger unemployment situation could beryllium looming.

“Younger workers consequence losing nan entry-level ‘stepping stone’ tasks that thief them build careers,” Mark Graham, professor of Internet Geography astatine nan Oxford Internet Institute, says.   

While AI enthusiasts admit that immoderate jobs will beryllium eliminated aliases automated, they will often reason that it will create much jobs than it destroys and free workers from doing elemental work. With much clip connected their hands, group tin attraction connected addressing higher-level problems successful their industries. Graham highlighted that this is not ever nan case.  

“In Amazon warehouses, for example, AI systems are utilized to way productivity infinitesimal by infinitesimal and automatically emblem clip disconnected task, which has raised concerns astir activity intensification and occupation quality,” he says. “Instead of freeing group to deliberation astatine a higher level, nan exertion location often narrows discretion and increases surveillance. It shows that whether AI empowers aliases constrains workers depends little connected nan instrumentality itself and much connected really employers take to deploy it.”

In a world wherever nan Stanford study reflects nan commencement of a structural problem successful younker unemployment, and not conscionable a short-term labour marketplace accommodation to a caller technology, nan domiciled of AI upskilling programmes whitethorn beryllium moreover much important.  

However, Graham argues that, successful this scenario, nan UK would besides request to grow redistributive policies. “The drawback is that astir of nan large winners from AI firms are based extracurricular of nan UK, which limits what Britain tin raise done firm taxes. That makes world taxation coordination essential,” he says. “Upskilling matters, but it can’t lick nan structural reality of less jobs.”  

What does successful AI upskilling look like?  

In August, MIT’s Networked Agents and Decentralised AI task recovered that 95% of generative AI aviator programmes are failing to thrust revenue, delivering small to nary alteration successful profit and loss. According to nan study, nan main driver of this was a “learning gap” for devices and organisations. The companies that were nan astir successful successful utilizing AI to thrust gross tended to acquisition outer AI devices alternatively than building soul ones and focused on automating back-end processes.  

So, what makes a bully AI upskilling initiative?  

“The top occurrence I person seen is wherever we commencement pinch a business result and activity backwards,” Woodstock notes. “It has been very mixed results erstwhile it is simply a generic ‘let’s get everyone to usage AI more’, and it has been transformational erstwhile location is an uplift that we want to spot successful a type of business consequence aliases a level of personalisation successful nan work that we want to deliver.”  

Townsend besides underlines nan value of focusing connected outcomes, peculiarly arsenic a measurement to show workers nan advantages AI tin bring them.  

“I deliberation really focusing connected nan circumstantial outcomes and really it is going to amended productivity will beryllium nan biggest alteration that we request to make arsenic we are communicating this retired to people,” he outlines.  

National upskilling initiatives should truthful beryllium underpinned by 1 question: what are they trying to achieve? Otherwise, they consequence falling into nan consequence of ‘generic’ take Woodstock warns about, wherever workers are told to usage much AI conscionable because. As nan AI hype dies down, if initiatives aren’t outcome-driven, they will besides neglect to impressment overseas investors.

Given nan accelerated gait of change, nan biggest instruction for governments and companies is that training programmes mightiness beryllium present to stay.

As Townsend says: “We can’t trust connected skills we study astatine 21 and past beryllium done. That is conscionable not really it useful now.”