Is the AI bubble about to burst?

Despite all the hype, AI may end up boosting productivity by just 0.5%, according to one MIT professor

While the HR hype around AI shows no sign of slowing, the volume of its critics is increasing. Dan Cave explores what HR really needs to know, and be mindful of, about AI.

Hype around artificial intelligence (AI) has exploded over the last couple of years. The World Economic Forum believes that AI will change around half the jobs in developed nations.

Many companies use AI internally and in products. Its popularity led to a chip-making company behind much of AI development briefly becoming the world’s most valuable company earlier this year.

This isn’t to say that AI’s rise has been inexorable. In recent months, many AI-powered firms have had billions wiped off their value.

According to an MIT university professor, AI will eventually only change 5% of work, and boost productivity by just 0.5%. Goldman Sachs researchers say that the tech will make existing features more glitzy than useful. Where do such doubts leave HR?

AI and HR

Many of the biggest businesses are already using AI software across the talent lifecycle. Hilton uses it for recruitment, PepsiCo for engagement and lifecycle management, and Shell uses it for learning and development.

Many in HR will recognise a version of AI being integrated into recruitment for years, initially to mitigate against unconscious bias and to streamline hiring. This integration expanded into multiple areas such as performance management, learning and development, career development and workflow optimisation.


Read more: The EU AI Act: Will UK employers be affected?


Already, the HR tech marketplace is awash with AI offerings that promise to save time and help HR deliver strategically. What’s not to like?

Truthfully, in many areas of HR it’s hard to see the AI hype fading. While the financial services company Klarna is looking at phasing out human employees to be replaced with AI, Tim Sharp, global head of talent at the pharmaceuticals company Takeda, explains that AI is used across the entirety of HR’s remit: from skills development to communicating meeting outcomes to employees.

“We’ve got it everywhere,” he says. “It’s really useful. It can maximise both outputs and operating models.”

HR’s belief in AI

Sharp isn’t the only one who sees AI as positive and practical. More than six in 10 HR leaders want to use AI more, research by insights firm Gartner found, and many see AI as nothing but a boon.

While the world at large might be starting to doubt AI, Gartner found that 81% of HR leaders use AI to, as they perceive it, boost efficiency. More than nine in 10 report believing that AI saves costs.

Neil Sahota, CEO of the machine learning software firm ASCIILabs, cites multiple places where AI is being used for good in HR: Unilever leverages AI tools in video interviews, aiming for more objective, unbiased assessments; Hilton uses AI chatbots to save practitioners time, and the multinational firm General Electric uses AI for continuous performance assessments. “AI is transforming the way organisations manage their workforce, offering innovative solutions that enhance efficiency, accuracy and decision-making across various HR functions,” Sahota says.

Elsewhere, Jon Lester, VP of HR tech and AI for outsourcing firm IBM HR, says that AI has allowed HR to dump repetitive tasks. “HR professionals are using the extra time to learn new skills,” he explains.

Shmulik Barel, VP of the software firm Workhuman, sees AI as giving HR the competitive edge in areas that have been traditionally hard to capture and measure, such as employee sentiment. He says: “AI helps HR make evidence-based decisions to support strategic goals.”

For those who might be tempted to buy into catastrophic assessments of where AI might take the world of work, or who are frustrated that it hasn’t delivered yet, global industry analyst Josh Bersin advises looking at the examples of where AI has already helped, such as data analysis and equality, diversity and inclusion efforts: “We have so many challenges in HR that are essentially data-driven problems, and we don’t have the time to parse it. AI can do this, and help us make better decisions along the way.”

AI worries

But AI doesn’t come without risk. Bersin admits that AI’s utility can be oversold, and that it is well known that generative AI can hallucinate (produce false, misleading or fabricated information). Vendors can get lost in the data rather than the solution HR practitioners need, he says, and, as it stands, many AI products focus on broad utility rather than acute HR problems: “The information that comes from this type of AI is likely to be B-minus level,” according to Bersin.

Sahota also sees limits to AI, adding that while AI is used to mitigate bias, it can also exaggerate bias, creating ethical, and operational challenges. “AI can perpetuate implicit bias if not carefully managed,” he says.

“We’ve seen examples of it penalising candidates from women’s colleges, which shows the importance of transparency and diversity in AI training data.” He adds that AI needs human oversight to maintain trust and humanity alongside its use.


Read more: HR has an AI-powered disability problem


Similarly, IBM’s Lester believes that AI is still lacking in some key areas. Specifically, he says, it is not up to speed in aspects of organisational life that are highly regulated, adding: “AI has limitations, particularly in areas requiring human empathy and nuanced decision-making.”

As professor and AI ethicist Roger Steare explains it, organisations implementing AI need to ask what it’s actually delivering, and if it’s benefiting humans. This is especially critical when metrics such as employees’ psychological safety, happiness and engagement are central to HR’s perceived performance.

Steare says: “Many people think AI is conscious or has a conscience. It doesn’t. As with much technology we introduce into work, we have to ask more: is it delivering productive value, or helping people lead happier lives?”

Getting past AI problems

Such questions – about how to manage AI effectively so that it benefits both the business and its people – will likely land with HR. So, Bersin says, HR professionals should look for vendors that can show workplace value, and that have a targeted solution.

For Lester, it’s about ignoring vendors that promise the world and instead looking at those that see AI as one tool among many. “AI should be seen as a powerful tool in our toolbox that augments HR processes rather than replaces human expertise,” he says.

Indeed, it’s the increasing tailoring of AI that Sahota says will lead to a more mature use. “Chasing after the next big AI breakthrough, many HR professionals are focusing on incremental improvements that can deliver tangible benefits.”

What does this mean for managing the technology? For Eser Rizaoglu, Gartner’s senior analyst, it’s about having a product owner, as with other parts of the tech stack. He says: “The key recommendation for HR leaders is to introduce an AI product owner role into the HR function who can not only coordinate HR’s approach to AI but also work with the broader organisation to ensure that AI has a positive impact on the whole workforce.”

Lester adds that it’s about getting the business case right, getting the right AI skills into HR (such as prompt proficiency), being mindful of drawbacks, and building out policy and ethical oversight, as well as being transparent about the data that AI uses. “For HR, the rule of ‘humans in the loop’ is key to ensuring that AI does not make decisions about people, and people make decisions about people,” he says.

As more HR leaders build AI into their processes, the rapid pace of transformation means oversight is needed, according to Andreea Wade, VP of AI at software firm iCIMS. “It needs robust governance,” she says, meaning ethical guidelines and policies that ensure AI is human-led, technically robust, private, secure, transparent and accountable.

As Stear recommends, some scepticism is probably warranted. “Scepticism is a healthy process: it’s not about being a Luddite but about weighing up the pros and cons of AI for your workplace,” he adds.

Where next?

As Bersin sees it, people will veer away from utopian promises as the AI market matures. The mindset of HR leaders will shift from a focus on how AI is changing the business HR is in, and move towards how AI is improving what the business already does.

HR is only at the beginning of its AI journey, Bersin explains. If vendors and buyers make the right choices, then AI might be applied to problems that need fixing. He adds: “Of course, this means taking the usual approach to implementing anything new: having an ROI measure, and ensuring people do need it.”

For now, according to Steare, HR should take a balanced view: questioning whether vendor promises are truly all that, while understanding where AI might have potential. “There are areas where AI is being used well, in conjunction with human failsafes. We can point to areas where it does lots of good.

“But there are dangers. We have to ask vendors to show the technology’s processes. We have to care about the value, the ethics, safety and morality,” which is never a bad thing for HR to do, anyway.


Read more: Uber Eats driver wins payout for racially biased AI checks


HR’s AI doubts

While many HR professionals want to use AI more, others are more sceptical about its use. Tina Rahman, founder of the consultancy HR Habitat, doesn’t want to use it when it comes to employment law and HR, cutting off contacts who kept trying to push it on her. “I was always against AI when it comes to employment law,” she says.

Laura Evans, CEO of the consultancy Glass Ceilings Change Management, believes that AI should be treated with more caution, especially with regard to intellectual property and personal information. “We will not use any external AI tools to transcribe meetings, and we ask others to switch theirs off before starting a conversation,” she explains.

The CIPD’s director of profession, David D’Souza, reminds us that HR’s needs may be different to vendors’. He says: “HR professionals should take great care when selecting a technology vendor, to ensure that the choice matches their needs and risk appetite.”

AI and HR chatbots

AI-powered chatbots are increasingly popular for businesses. They promise to streamline processes, personalise tasks and create recommendations. Many integrate into other tech stack elements, and can be managed using voice commands.

“Our solution has been particularly popular in companies with complex HR needs and global workforces,” says Michelle Dawkins, senior director of solution consulting for the AI-powered HR tech provider Workday. “It enables HR professionals to dedicate more time to strategic initiatives.”

Gartner research has found that employee-facing chatbots are among the top three uses of generative AI in the HR function. Such bots are now akin to digital assistants that automate tasks and answer FAQs.

 

This article was published in the September/October 2024 edition of HR magazine.

Subscribe today to have our latest articles delivered to your desk.