Workers actually have significant leverage on AI right now
Businesses are desperate for workers to adopt AI, but what offer are they making?
This is a guest post from Adam Cantwell-Corn, AI policy lead, Trades Union Congress
Image credit: Yutong Liu & Digit / Better Images of AI / CC by 4.0
In talk of the ‘AI transformation’ workers are often cast as the subjects or even victims of change. They are to be trained, maybe protected or perhaps just discarded. It is true that workers face instability. But two dynamics are playing out in a way that is giving workers, representatives and allies significant leverage right now.
First, while AI has the promise to automate or streamline tasks and to extend the capabilities of workers, adoption is weak and transformative productivity has not materialised. The white-collar bloodbath threatened (or promised) as machines take care of whole new task bundles has not arrived. CEOs are reconsidering AI usage. Rachel Reeves set a target to make the UK the G7’s fastest AI adopter but just 1 in 6 UK firms are taking up AI. Weak returns mean a bubble looms large.
Second, AI is fast becoming an electoral issue. In the USA, the aggressive data centre build-out has triggered a bipartisan ‘tech-lash’. OpenAI dangling the jaded promise of a shorter working week in its ‘people first’ AI industrial strategy reflects Silicon Valley’s concern that the public is turning on them. Anti-AI populists are cornering Trump in the mid-term elections.
In the UK the signs are emerging too. Seven in 10 of the UK public are worried about the economic impacts of AI. Protests against data centres are gathering steam inside and outside of Parliament. And Palantir’s NHS contracts may be ditched after workers and the public mobilised against it.
For different reasons Big Tech, business and government are all keen to drive AI adoption and head off a backlash - and they need workers buy-in at the ballot box and in the workplace. This gives workers a window of leverage to demand a better offer. What might that that look like?
Workers’ experience and expertise is leverage to demand a better offer on AI
Let’s start with the current reality in workplaces where generative AI could have the most impact. In certain settings AI clearly has value, but beyond software engineering its use is not yet systematised. Many workers are quietly using AI to keep up with and complete tasks. But this is ‘shadow IT’, hidden from employers for fear of being given yet more to do. Some employers are monitoring employees to make sure they meet usage targets, resulting in ‘adoption theatre’.
These failures are driving the recognition that as with all other technological transformations, meaningful innovation requires organisational development, process redesign, and upskilling.
As the CMI found, bosses alone are not equipped to do this, or to sift through the snake oil. Doing so requires respecting the workforce’s expertise, experience and interests to effectively deploy technology. As a case in point, car maker Ford recently rehired veteran engineers due to failures in AI systems, with an executive saying “we didn’t pay as much attention as we should have to the experience of our most knowledgeable engineers.” Even Microsoft has cottoned on, saying that ‘organizational AI adoption depends on employees as much as leaders’ and that ‘centering worker voice in AI design boosts productivity, satisfaction, and skill growth’. Despite a long-period where worker’s autonomy and influence on the labour process has been degraded (as per Harry Braverman’s thesis), workers know the pain points and the realities of workplaces in a way that bosses do not and cannot.
Though exposed to disruption, workers in sectors where generative AI is most applicable have leverage while organisational innovation remains the bottleneck. They can demand that employers step out of the hype and determine whether and how AI can help with a defined business problem (it might not). Workers can push the conversation about improving job design, training and the need to address the evidence that AI can lead to de-skilling, intensification and ‘work-slop’.
There are positive use cases. But in each case, it’s the governance of work that matters as much as the technology. Social workers in the UK report high satisfaction with voice-text transcription software – but concerns exist on clarifying how skills will be protected and liability for AI hallucinations managed. Workers in logistics are keen advocates of health and safety applications - but want to ensure that inward facing cameras in lorries are not primarily used to surveil and sanction workers.
Most fundamentally, workers and their unions can push to negotiate AI adoption so that any gains are shared – as higher wages, improved work or shorter working time - and not spirited away in profits or used as a rationale for job cuts. This is hard work, especially for workers on precarious terms of employment.
For employers keen on AI but allergic to this ‘red tape’ it’s not coincidental that the Nordic countries have among the highest levels of AI adoption. Scandinavian workers have more power to shape technology and are therefore more open to its effective use. Visiting unions in Stockholm recently I was struck by what one trade unionist said: “We don’t fear new technology, we fear old technology”. He was confident that social democratic institutions such as sectoral bargaining would ensure workers could gain from innovation and be supported through disruption.
Workers need a transformative offer to accept ‘transformative AI’
Which brings us to the need for the government to step up an agenda for a pro-worker AI economy, and to tilt the balance towards workers. Recent months have seen a shift to address the souring politics and weak AI adoption.
The TUC participated in the ‘AI Adoption Summit’ in June, which saw ministers highlighting the essential role of unions and a worker-centric approaches to AI, along with the launch of initiatives on early careers and the AI Economics Institute.
But a tack in narrative – while a good step – won’t be enough to avoid a backlash on AI. If the government wants workers to embrace AI ‘transformation’ then it needs to make an offer that is equally transformational. Without this it is quite rational to push back. There is no durable and progressive political basis for AI adoption when 65% of Brits think AI gains will be mostly captured by large firms and investors.
It is true that policy action is tough amid uncertainty of how AI may play out. However, workers can demand that government adopt a ‘no-regrets’ approach that would be positive in any eventuality.
These include a shift to taxing wealth fairly compared to work. It means a legal right to shape high-impact AI at work, and reforming corporate practices so that workers are stakeholders in firms, not merely inputs to be minimised.
Going wider, the state should decisively mobilise its vast spending power to gradually rebuild our sovereign technological capacity while enshrining good work standards in all providers and taking on – not courting - toxic tech firms. Crucially, the government needs to fully implement the landmark Employment Rights Act to empower unions to establish collective bargaining in highly exposed sectors.
Workers should use the power they have now to secure a voice, rights and a fair share on AI at work. And business and government need to step up with a better offer. As well as material issues, having a stake in changes in your workplace and the wider economy is a matter of dignity – so important for our country’s tired social fabric. Without this, AI disruption – real and perceived – will become ripe for exploitation by the far-right, risking the wheels of technological and social progress coming off altogether.