Generative AI has long been hyped as a technology which could ‘add value’ to businesses, while simultaneously being billed as a catastrophic force that could render swathes of the workforce redundant. A new study from Sia Partners suggests the technology could see a third of back office jobs lost in the future – but as AI publicly struggles to actually delivering on its promises, critics are also suggesting that the hype might be a cover for much more traditional labour practices.

A new study by consultancy Sia Partners suggests that artificial intelligence, and the rapid expansion of large language models (LLMs) and GenAI technologies, are poised to transform how office work will be performed. Scanning 1,000 work-related tasks across 50 occupations, including hospitality, healthcare, engineering, manufacturing, legal, education, finance and the creative industries, the study found that the vast majority of roles will be impacted by the technology in some way.

The largest portion of tasks will see fairly minimal changes. According to Sia Partners, around 79% of occupations could see roughly one-tenth of their tasks influenced by the adoption of GenAI. This would most likely see tasks ‘accelerated’ by GenAI and LLMs, rather than seeing wide-spread job-losses in relevant sectors.

However, Sia Partners found that the future might be less rosy for workers in 28% of occupations. According to the study, 80% or more of tasks for this segment of the workforce could be exposed to the promises of GenAI – leading to a reduction in headcount, as bosses look to manage their bottom-lines.

Source: SiaPartners

The researchers noted that jobs in tax, web administration, IT and digital design were “the most heavily exposed”. And while little has perceptibly changed for the workforce since the promises of AI exploded onto the scene two years ago, the report insists this is only a matter of time – as soon as companies find ways to develop clear upskilling or reskilling strategies, helping workers get the most from GenAI. At present, most organisations lack a comprehensive approach to assess their workforce’s readiness.

“The key insights of this study highlight the transformative potential of GenAI, demonstrating the benefits it offers and the challenges it poses, while emphasising the need for strategic planning and thoughtful implementation,” said Andrei Pascanean, senior consultant data science at Sia Partners.

But advocates of AI may face greater challenges, than simply ensuring the right training is in place to make the most of the technology. Increasingly, hype-fatigue seems to be plaguing the so-called AI revolution, as the companies powering it struggle to move beyond the ‘hallucinations’ (known in the human world of work as mistakes) the technology was prone to 24 months ago – not to mention the huge costs involved in the process.

To many, it remains unclear whether generative AI truly provides much business value. Earlier in September, The Information reported that customers of Microsoft’s 365 suite (including Word, Excel, PowerPoint and Outlook, as well as a number of business-focused software packages which in turn feed into consultancy services from Microsoft) are scarcely adopting its AI-powered “Copilot” products. Only between 0.1% and 1% of its 440 million seats are paying for the features, which essentially double the annual cost of the license.

That report quoted one firm testing the AI features, which suggested “most people don’t find it that valuable right now”. Another added that “many businesses haven’t seen [breakthroughs] in productivity and other benefits” and they’re “not sure when they will.”

That is not to say there is no cull of back office functions on the cards. A number of global firms have slashed back office functions in recent months – and some have even done so while touting new ‘AI’ capabilities to pick up the slack from those losses. But if most firms are not actually using AI to fulfil the work, who is actually doing it?

One option seems to be that the roles are set to be outsourced to cheaper labour markets – a historic tradition under global capitalism, which is both less ‘revolutionary’, and less palatable to public sentiment. But with several market shocks threatening to stymie the record rates of funding AI developers still need to improve the technology in a way that fulfils on its promises, using AI hype as a smokescreen for farming out jobs to low-wage economies might be the most stable future that AI actually has.