It’s a subject that attracts endless debate: will AI replace whole swathes of human jobs and create an exodus from white collar professions?
The trigger for me writing some reflections on this question is that no other than Jeff Bezos weighed in on the topic recently when speaking at a conference in Paris. His view was that far from decimating human roles, AI would unlock new opportunities and increase demand for human labour.
I am in the Jeff Bezos camp. My belief is that just as other defining shifts like the creation of the internet created whole new classes of employment, so too will AI. That is not to say that it won’t redesign or remove some roles as the technology takes over certain tasks and processes itself – but it will also spawn new human activities.
As an example of how things are changing, take software development. Increasingly, and as is already happening, AI will write new code itself. But skilled human engineers are still needed to manage the process, and oversee and check the results. Crucially, their human skills and experience will continue to be essential in making sure that the products being developed meet the needs of customers and support the strategic priorities of the business. That human understanding of the wider picture is something that AI may never be able to replace.
An ‘AI boomerang’?
As in any transformation, there will be something of a rebalancing. Interestingly, we are already seeing signs of this in the market – back in the favour of human skills. Talk is growing of an ‘AI boomerang’ amongst some organisations who cut back on existing staff or new hiring due to AI, but who are now beginning to push the balance back.
There are several reasons for this. One is that it is still very early days in the AI revolution and organisations are still trying to find the appropriate model to set themselves up for it correctly. Another is that many businesses are still struggling to achieve real ROI from their AI investments – partly because it is early days – as our own research at Harvey Nash has shown: the 2025 Digital Leadership Report (DLR) found that only a third of businesses have achieved demonstrable ROI. A third element, related to this, is that AI is proving expensive. The newer AI models consume large numbers of tokens – the ‘burn’ is significant. These factors are all contributing to that boomerang effect. More broadly, the US technology recruitment market is holding up, with certain roles (AI engineers, data scientists, DevOps professionals, cybersecurity specialists) in high demand.
AI will be additive
There are other reasons to be optimistic from a human point of view. Our DLR also finds that the more an organisation invests in AI, the more likely they are to recruit technologists, rather than reduce the size of their workforce. In fact, we found that those organisations most ahead with large-scale implementations of AI are 24% more likely to be increasing their tech headcount than their peers, mostly in areas of AI and data.
Other studies bear this out. For example, the Financial Times reported recently on research by Ramp across almost 22,000 companies in the US which found that firms making the largest AI investments grow employment by roughly 10% following adoption, while low-intensity adopters see no statistically significant change. Entry-level headcount rises 12% for high-intensity adopters too.
This fuels my belief that AI will overall prove additive rather than taking away. It will create changes in how people work and what they do, but used effectively it will be an augmenter. The task in front of organisations is to get the balance right. It will be rather like seasoning a dish: you don’t want to go too heavily with any one ingredient but rather bring out the flavours from the right combination.
Upskilling is crucial
However, there is another element that is key to getting this right: upskilling. Whether it’s the tech team or other staff, it is critically important that people are empowered and enabled to use AI tools effectively, getting the most out of their capabilities and also managing the outputs appropriately. But this is an area where many businesses have more to do. The Harvey Nash Tech Talent & Salary Report found that while the majority of technologists (76%) have access to AI tools, far fewer feel they are being given structured support, time or clear frameworks to build capability (36%). Less than half (43%) are being given dedicated time to experiment and learn. A significant proportion report being expected to self‑learn (20%), while others are still waiting for formal training to be introduced (23%).
I would urge organisations to step up their efforts in this area. Equipping people for AI is the real battleground in front of us. Investing in learning & development programmes, giving staff time to experiment and learn, unlocking knowledge transfer via external parties the business works with such as consultancies – all of these will contribute to the goal. It is worth exploring governmental support too. For example, in the US, the Department of Commerce’s Economic Development Administration (EDA) has launched a new initiative to put $25 million towards the AI Upskill Accelerator, a national grant competition to fund industry-led sectoral workforce partnerships that help upskill and reskill workers around AI technologies. Similarly, in the UK, the government is tackling the national AI skills gap through the AI Skills Boost, which aims to upskill 10 million workers by 2030. Led by DSIT and Skills England, the strategy combines free online training with major tech firms and reforms to technical education.
Opportunity ahead
The impact of AI on human jobs is still unfolding, and it will be a nuanced picture. However, I take a positive view. There seems little doubt that the organisations who harness AI effectively will be the winners of the future, launching value-adding new products and services, finding ways to do things smarter, boosting productivity. This brings enormous opportunity – and people will be at the heart of making it happen.
