How much is AI really changing the need for skills?

It's true that some jobs - and particularly some skill sets - are being displaced by technology, particularly AI-driven automation. But this has been the case throughout history, and the key to managing such changes in the labour market is skills. That was the gist of the conversation at the Global Lifelong Learning Summit (GLLS 2024) in Singapore earlier this week.

People Matters was at GLLS, and picked up some insights from keynote speakers and panelists on the topic. To begin with, the change brought by AI is one that will - ideally - push the workforce towards higher skill levels.

More AI leads to more high-skilled jobs: OECD data

Delivering an opening presentation at the Global Lifelong Learning Summit, OECD senior economist Stijn Broecke commented that even though almost one-third of jobs are at high risk of automation, employment rates are still at record highs in the majority of OECD countries.

This, he explained, is because AI creates jobs both directly and indirectly. The jobs directly created are a very small handful of highly specialised roles focusing on developing and maintaining AI technologies - but OECD data shows that noticeably more jobs are indirectly created.

"The more exposed an occupation is to artificial intelligence, the higher the employment growth has been," Broecke said. But the caveat, he pointed out, is that high-skilled occupations, which have a higher share of non-automatable skills, are far more likely to see employment growth than low-skilled occupations. And almost all OECD countries, he said, have already seen a fall in the number of jobs for low-skilled workers even as jobs for high-skilled workers increase.

What kind of skills is AI replacing?

Early data from OECD observations suggests that white-collar support skills are among the first to be replaced. Broecke listed business and management skills such as project management skills, budgeting, accounting, clerical administration skills, basic work processing and spreadsheet skills, the basic capabilities required for jobs in finance, HR, legal, communications, administrative assistants, project managers, and similar roles.

"I don't want to say that these jobs are going to disappear, but what seems to be happening is that the kind of skills demanded in these jobs seem to be changing," he said. "This is preliminary evidence, but it's something to keep an eye on."

But AI isn't just replacing skills or pushing workers to change their skill sets, he added. It's also narrowing the gap between the least efficient workers and the most efficient workers, according to data showing that in occupations more exposed to AI, those who are less productive catch up with those who are more productive - and in these occupations, there is also less wage inequality.

The implications of AI-driven change on skills policies

What this means for individuals and institutions is a shift in focus. But it won't be a shift towards specialised roles or technical skills, according to Singapore's Minister of Education Chan Chun Sing. In his speech at GLLS, he said that individual workers or learners have a personal responsibility to stay relevant: not through digital skills, but by learning how to use AI responsibly and ethically, which means having a basic understanding of how AI models work and what their biases and limitations may be.

"AI cannot help us to discern," he warned. "It can distil information very well, but it cannot discern so well because that requires human judgment and value systems. It can discover and create new value propositions, but not everything."

And - contrary to the interest in technical AI skills that burgeoned when generative AI first came on the scene - he highlighted soft skills and mindsets as being the most important focus.

"Some things will not change," Chan said. "Values and interpersonal relationships will continue to define how we work and learn, and we should be careful not to over focus on digital and AI upskilling at the expense of other critical core skills such as critical and creative thinking as well as collaboration. Our lifelong learning must continue to be fueled by a spirit of inquiry."

Is AI going to take your job if you don't skill yourself correctly?

No, said the GLLS speakers - but your colleague who has learned to use AI might take your job. In an international panel dialogue on the changing demand for skills, Emily Jefferis, head of audit quality at KPMG UK, went into detail on the point about AI being able to level up employee performance.

Using the example of how auditors' performance is assessed by the quality of their written work and their ability to provide supporting evidence, she said that AI tools are now able to give feedback and coaching on how to improve those areas, effectively closing the gap in performance.

"So how do our most productive employees continue to differentiate themselves?" Jefferis put her finger on the crucial question. "That's with the focus on the most critical human skills: the ability to build relationships with the entities that we are auditing, to be able to effectively coach their team members, to be able to ask really critically challenging questions."

And even the digital skills that workers will need in the future are going to be more about how they as humans interact with and use the AI, rather than the technology itself. David Timis, global communications and public affairs manager at global employment nonprofit Generation, took a few moments out of the same panel to mythbust.

"It's a common misconception that with digital skills, you need to learn how to code," he said. "No. You just need the basics. It's the same with AI. We're not teaching our learners at Generation to code. We're just teaching them the basics of AI literacy, which have been mentioned by my co-panelists. The most important is prompting. And I would myth-bust again: prompt engineer is not the job of the future as the media makes you believe. Prompting is the skill of the future. We will not need millions of prompt engineers, but we will need hundreds of millions of people who know how to ask these tools the right questions, because the better the question, the better the result you get."