Like many social issues in recent years, the increased availability and use of artificial intelligence (AI) appears to be a polarising topic. For some, it is seen as a great mark of progress, and for others a potential disaster, robbing us of human influence in so many tasks.
This is particularly true in relation to the application of AI in the workplace and within HR. On the one hand, AI could potentially increase efficiency and reduce the risk of human error. On the other, any reliance upon AI appears to go against the nature and function of the role – Could it risk taking the ‘human’ out of human resources?
The use of AI in the workplace
In reality, AI (essentially, the application of an algorithm to recognise patterns in data and make conclusions to ‘automate’ certain tasks) has already been in operation within the workplace for many years.
Most email and IT security systems use AI, as do many employee monitoring systems. Many larger companies use AI to record attendance and performance of employees and to trigger relevant formal procedures as well. These tasks, which might otherwise require members of the HR team to input and then study reams of data, can now all be automated. So too can the production of the various management reports that might be informed by the data.
Some employers go further still and use programmes such as ChatGPT to generate material, such as answers to frequently asked questions from employees.
In reality, almost everyone now uses AI to help them in either their work or daily lives; who can honestly admit that they have not resorted to Google when trying to answer a tricky question?
By and large, the benefits of AI are clear; the improved efficiencies and elimination of human error all serve the overall advantage of saving costs.
Potential legal issues with AI
There are, of course, potential problems and risks involved with the use of AI. Generally speaking, AI relies upon an algorithm having access to significant amounts of data. Therefore, there are potential issues surrounding data protection.
Without meaning to conform to lazy stereotypes, there are also genuine concerns that employers could become too reliant upon AI and that they get trapped into a “computer says no” style of employee management. This could undermine the key feature of trust and confidence in the employee relationship, and could also lead to claims of discrimination.
As with many things, it is when things go wrong that AI can gain significant media attention. Many people may be aware that employers famous for their use of technology, such as Uber Eats and Amazon, were themselves subject to claims of discrimination based on their use of AI.
AI shone a light on Amazon’s pre-existing bias
In the case of Amazon, a recruitment tool studied applications the business had received over a period of ten years to then screen future applications for the best candidates. But not unlike many businesses in the 2000s to 2010, a high percentage of applications had been from men, and a number of senior roles were won by and occupied by male candidates. As a result, the tool effectively inherited the pre-existing bias that had existed within the business, by teaching itself that male candidates were preferable to female.
While cases such as this highlight some of the serious risks involved, they need not be a warning against the use of AI. Instead, they can serve as a useful reminder of some of the precautions that employers might exercise to ensure that its use can be safe and successful.
(Read about how to close the AI gender gap)
Practical steps to manage AI risks
There are a number of steps employers can take to mitigate the risks of integrating AI tools within your workplace.
Risk assessments
Given the relationship between data and AI, one of the most practical steps that employers can take when managing the use of AI is just the same as any issue concerning the collection and processing of personal data; complete a risk assessment.
This should be done before adopting any form of AI or automated management systems so that the employer can assess where those risks lie, and how they might be reduced or eradicated.
The UK Government does not yet appear to have any plans to introduce any new form of regulation concerning the use of AI.
Protected characteristics
In relation to the use of personal data, employers must ensure that employees are aware of what data the system might be collecting and why, and then how it will be used by the business.
Employers should also consider whether the use of any AI tool might impact upon certain protected characteristics. Is the tool able to recognise those characteristics, and does the input data represent a sufficiently diverse sample of the existing or potential workforce?
For example, could a tool measuring attendance recognise that an employee may have a disability that causes a greater level of absence from work? Or could a voice recognition tool, which may be used to help with efficiency or security recognise all the different accents amongst the employees? If not, then a business may be exposed to claims of discrimination or harassment.
AI policies
Even when employers feel confident they have addressed risks such as these, the process of implementing AI tools still requires management. Ideally, leaders should engage in consultation with employees and then develop policies which explain what the tool will be used for, and how it might be regulated.
AI training
Most importantly, however, employers should ensure that they train HR teams and managers to understand the tools used, how to use them ethically, and how the results it produces might be interpreted.
This is key, especially as the UK Government does not yet appear to have any plans to introduce any new form of regulation concerning the use of AI. So far, it is relying upon the existing laws that regulate the employment relationship.
Ultimately, the key role remains with the humans in the HR team to continue to manage and safeguard both the business and their employees.
