While AI has the potential to create immense value for businesses, the key to unlocking this is finding suitable applications and use cases for the technology – which not all companies have been able to do.
Offices devoid of human workers. Streets occupied entirely by driverless vehicles. Factories and restaurants running on automation alone. While these may be extreme scenarios, artificial intelligence (AI) tools introduced in recent years are disrupting labour in nearly all industries. How is AI shaping the workforce, and what will be its effect on the human labour market?
This topic took centre stage at a Tech Talk X webinar organised by digital@INSEAD. Moderator Vikas A. Aggarwal, Associate Professor of Entrepreneurship and Family Enterprise and Academic Director of the Global Private Equity Initiative at INSEAD, was joined by industry experts Burcu Bicakci (MBA ’07D), Partner at Egon Zehnder; Fernando Lucini, Global Data Science & Machine Learning Engineering Lead at Accenture; and Karli Kalpala, Head of Strategic Transformation & Financial Services Industry at Digital Workforce Services.
Opportunities and challenges of AI
Aggarwal began by elaborating on his research, which he conducted with INSEAD’s Ivana Naumovska and Erdem Dogukan Yilmaz from Erasmus University. Their study investigated the substitution effect of AI on human labour, specifically related to knowledge workers.
They found that the launch of Google Translate led to a significant reduction in human translation transactions, as well as a decline in the earnings of translation workers. But transcreations – which involve adapting meaning to cultural and emotional elements in language, such as in the fields of marketing and poetry – remained relatively automation-resistant and did not experience a drop. The researchers also observed a significantly negative effect on the demand for analytical content after ChatGPT was launched, while creative content seemed to be less affected.
“What we’re highlighting here is a nuanced AI substitution effect across different types of tasks within a given profession of knowledge workers. This suggests that it’s necessary to understand variations in the AI substitution effect and [how] firms can adapt their workforce accordingly,” Aggarwal said.
While AI has the potential to create immense value for businesses, the key to unlocking this is finding suitable applications and use cases for the technology – which not all companies have been able to do. Kalpala remarked that the challenge isn’t typically the technology itself. Rather, it’s identifying large enough chunks of activities within an organisation that are worth the effort to transform into an AI-driven capability.
“There has been a massive move towards generative AI use cases… but is it really materially changing things? Not quite yet,” said Lucini, who added that using AI as a co-pilot is one of the most popular uses of the technology today. “If you’re doing software engineering, you may not want this thing to give you all the answers, but boy is it nice to have a sparring partner that is eternally patient.”
The role of leaders and organisations
Aggarwal highlighted the importance of AI to leaders, be it as a tool to inform decision-making or support how they innovate within changing organisations. Bicakci added that other than productivity benefits, AI can accommodate and close skill gaps, which can help foster greater workplace inclusivity.
Leaders who are not up to speed with using AI and similar data-driven tools risk falling behind. It is therefore imperative that they build the right skills and competencies to determine how to best employ the technology in relation to their existing workforce and business objectives. However, according to Bicakci, the speed of AI growth currently exceeds the rate at which this upskilling is taking place.
Additionally, leaders need to think about how they are approaching the relationship between AI and human labour. “AI does [not do] labour. We do labour, and AI supports labour,” stressed Lucini. “We need to understand [the role of AI] and integrate it into our lives in an intelligent and useful way, but never forget that it is a tool – let’s not assign intelligence or agency to it.”
The panellists pointed to confidentiality, trust, a lack of existing benchmarks and governance issues as some of the main hurdles that businesses must overcome to leverage AI productively. These problems will be magnified once the technology is adopted on a larger scale beyond the current “curiosity” phase.
“There is [presently] no governance in companies about how to use AI… which is actually quite a risky thing. The governors don’t exactly know what the potential is or where they need to be governing… and companies will probably do more than the regulators initially,” Bicakci said.
Looking to the future, governments – and, although less so, large companies – are the most likely to create their own generative AI models to overcome the security and reliability issues associated with commercially available options like ChatGPT. However, this should only be done after a careful cost-benefit analysis.
The panellists shared that employees should prioritise learning agility and upskilling to prepare themselves for greater AI adoption in the workplace. This could range from getting generative AI training to taking a coding class. Bicakci also emphasised the importance of raising one’s technology quotient – an understanding of and openness to new technologies.
She also stressed that many organisations are currently not ready to maximise the use of AI within their organisations, bar a small subset of firms that have already created the necessary infrastructure. “[Many firms] want to leapfrog into an era [of AI]… but the maturity level of the organisation is very much determining if they can build a good talent strategy [around it],” she said.
Lucini echoed these views. “Everybody has become an amateur in AI in the last year. But the truth is, nobody wants amateurs to build the next thing,” he said. “Not only do [they] want you to have done [something with AI] ten times, [they] also want you to be able to explain it to the CEO and to the organisation in a way they can understand… to help navigate the journey.”
Much has been said about how AI may benefit a few at the expense of many, with corporations enjoying the spoils while employees – in particular, blue-collar workers – experience greater inequality.
“I would definitely love to see a situation [in which] the benefits are spread around a bit more into society… than we are currently seeing,” said Kalpala. Bicakci expressed that while this may happen down the line, the benefits of AI are unlikely to be doled out equally within the next five to ten years.
“The benefits won’t be shared widely, it’s going to be with the elites,” she said. “Knowledge workers [will become] more powerful than ever… and low-skilled labour has the potential to be substituted in the short term. That means inequality [will be] coming into the picture.”
Aggarwal asked whether businesses should be tasked with mitigating this looming inequality within their organisations. For Bicakci, the responsibility does not ultimately lie with users. Although regulators have a role to play, she emphasised that it is ultimately up to AI developers to consider the broad social impact of their inventions before bringing them to market.
“Maybe you find a cure to automate everything in the [factory assembly line]. But you don’t launch it now because you don’t have a solution for the millions and millions of people that are going to lose their jobs,” she said. If you launch it tomorrow, then you create a big tsunami of people [who are] unemployed [and] social trauma.”
Lucini added that despite the widespread availability of AI tools, not everyone has the baseline knowledge and skills to make full use of them. “The massive exponential growth [of AI] does nothing but build a distance between those that can and those that cannot… the world has to catch up, or we’re all in trouble,” he said. “The rising tide will raise all boats, as long as you have a boat.”