In the three years since ChatGPT launched, generative AI has dominated headlines and reshaped how we think about work and technology.
One vision predicts mass job loss, hyper-surveillance, shrinking autonomy and massive environmental costs.
Another imagines AI easing workloads, improving service quality and speeding up innovation and productivity.
But, as many see it, future we will not be determined by the technology alone but by who has the power to shape it.
“When employers and unions sit at the table early — before systems are rolled out — we avoid problems later,” Oliver Roethig, a trade union leader at UNI Europa, the European federation of unions in private services, explains.
“We avoid fear, we avoid resistance, we avoid the sense that ‘this is being done to us’. Instead, workers can shape how AI is adapted in a given company.”
He says that many workers testify to the two-faced nature of generative AI in the workplace: “First, they see the potential. AI can simplify tasks, predict problems and support better service. Second, they worry about losing control — over data, over decisions, over skills.”
New predictions about AI’s impact on the labour market emerge every week.
A recent study by the German government predicts a disruption of 120,000 jobs. The Guardian predicts AI will replace three million lower-skilled UK jobs by 2035. And MIT’s detailed analysis concludes that AI could eliminate 11.7 percent of US jobs.
But almost none of the recently published studies show evidence yet of job loss because of AI replacement. For now, AI’s impact on the labour market is about transformation, not replacement.
Few know this better than workers and employers in the telecommunications sector, which maintains much of the physical infrastructure powering AI while applying the technology in network design, cybersecurity and customer service.
The Italian telecoms giant TIM, for instance, uses generative AI to transcribe and analyse customer care calls to anticipate problems and improve service.
As its vice-president of artificial intelligence, Clara Fabiola Oliva, explains TIM has also generated five “digital twin” customer profiles based on collected data to be able to test new products without needing to call their customers.
A recent joint statement between European trade union UNI Europa and telecom employers association Connect Europe acknowledges such beneficial potential of artificial intelligence.
But the organisations note that the “European Commission’s focus on innovation, investment and the potential productivity gains of AI alone is insufficient to ensure a sustainable and beneficial AI adoption.”
They insist on upholding European ethical norms, legal requirements and fundamental rights to build an "ecosystem of trust" in which Artificial Intelligence can flourish.
This includes a “human-in-command based use of AI” to avoid discrimination and misuse and stronger “social dialogue and collective bargaining” when new technologies are introduced at the workplace.
This is critical with emerging risks like “wireless hallucinations,” where AI generates inaccurate or illegal information, potentially degrading services, wasting resources and creating security gaps.
Sector analyst Rebecca Davies of Visionary Analytics warns that “the blame for such failures could be shifted onto workers”.
She also thinks the benefits of generative AI — boosting automation, enabling new digital offerings and promising productivity gains — remain “uneven and unproven at scale.”
The overstated expectations could risk burdening workers and eroding trust. Furthermore, unequal access to training threatens to widen divides, leaving older workers, women and employees of smaller firms most vulnerable.
That’s why Alessandro Gropelli, director general of Connect Europe, believes the sector needs more investment to train workers for AI: “Telecom networks are a crucial part of AI infrastructure. We need a better investment environment to enable telecom companies to invest more in skilling and upskilling their workforce.”
These discussions about training and ethical use are not as abstract as they may sound.
Across Europe, unions and companies have already negotiated ground-breaking agreements on AI use across the broader ICT industries.
Italian unions, for example, negotiated an agreement with contact centre company Konecta Italia.
Its key provisions, as Bianca Catapano from the union Fistel-CISL outlines, mandate that workers are informed before any AI deployment, which includes a 12-month experimental implementation period followed by an evaluation.
A joint committee, composed equally of worker and employer representatives, meets every three months to propose improvements.
Crucially, the agreement prohibits the use of AI for worker discipline or merit assessment, establishes training programmes and allows for union monitoring of the technology’s impact.
As a result of generative AI, contact centre workers — in the telecoms industry and beyond — will likely transition from direct task execution to oversight and verification.
Across the economy, workers will move into higher-skill roles, such as “data stewards,” overseeing AI processes rather than being replaced by them.
But only meaningful social dialogue between employers and workers can build the trust essential for such a transformation.
As Birte Dedden, director of UNI Europa’s ICT sector, insists: “If the workers do not trust the tool or the technology, there won’t be an efficient adoption and take-up."
UNI Europa is the voice of seven million services workers coming together from 242 national trade unions in 50 countries, representing sectors that constitute the backbone of economic and social life in Europe.
UNI Europa is the voice of seven million services workers coming together from 242 national trade unions in 50 countries, representing sectors that constitute the backbone of economic and social life in Europe.