HR and IT departments need to work more closely together if they are to invest wisely in artificial intelligence tools, according to an HR futurist.
Lydia Wu, who has held roles in Panasonic, Deloitte and Accenture and now consults with companies on digital infrastructure, said it can be tempting for HR departments to “not want to deal with AI and throw it over the wall to IT” because it feels too complex, but their role is crucial if AI is to be a success.
Speaking at the HR Technologies conference this week, Wu reflected on the rapid growth of AI tools in HR, particularly since the explosion of generative AI last year.
“Since the pandemic we’ve seen the mass consumerisation of AI and investment in computing power. Machine learning means tools can break things down to a deeper layer, look at what a human-type reaction would be to something,” she said. “But AI should never make a decision on a human’s behalf.”
Artificial intelligence in HR
AI for HR: What is it and how will it change?
This is why the role of HR in AI technology purchasing decisions is so crucial, she added. “HR has to look at the intent of the tool, how it allows a human to intervene and guide the last step of a process, how it ensures every answer can be human-verified, and cannot be replicated.”
Wu’s advice comes as the TUC has warned that the UK is failing to keep up with the regulation of AI in the workplace, leaving employees vulnerable to exploitation and discrimination.
Last year it launched a taskforce with the aim of drafting new legal protections for workers and their employers.
The TUC has proposed that employers should have a legal duty to consult with employers on the use of “high-risk and intrusive” forms of technology. Other countries have already introduced new laws on AI’s use.
Wu said HR professionals should ask five questions when looking to invest in AI HR tools.
- What model does the solution use? Whether a solution is closed or open-source will mean different levels of security risk, she explained.
- How is the model trained? HR needs to be part of discussions on the data that will ‘train’ the system to make decisions, and how much control and transparency there will be over it.
- How is customer data used? There may be data protection and privacy issues.
- Are the outcomes explainable? HR needs to consider the layers the system passes through before the decision reaches a human.
- How do we ensure the outcomes are verified and repeatable? If an outcome is verified, the human is at the centre. Being able to repeat the outcome is useful, but this is where bias can creep in, Wu added.
Any boundaries that currently exist between HR and IT need to be broken down to ensure there is a people-driven element in AI investment, Wu concluded.
“The wall doesn’t exist anymore. Every HR leader needs to be a tech leader as we move into the future,” she said. “There has to be a level of interaction.”
Sign up to our weekly round-up of HR news and guidance
Receive the Personnel Today Direct e-newsletter every Wednesday
