The introduction of AI systems promises beneficial changes for the workplace. As a result, businesses are becoming increasingly more receptive to AI tools to harness the benefits they bring. However, there are potential risks if AI is not used responsibly, writes Pam Loch of Loch Associates Group

Given the significant impact upon businesses – good or bad – AI should not be perceived as just an “IT” or compliance issue. Instead, it should be treated as a core “HR” issue. After all, HR would usually be dealing with any misuse or breaches that occur in the workplace and managing human resource; so does it not make sense for this function to be leading from the front and recognising the crucial role that HR plays in ensuring the responsible application of AI?

Risks and claims

While there are potential business benefits to using AI, such as increasing efficiency and boosting productivity and wellbeing, often there is a lack of understanding about AI and how organisations can harness this tool to reap the rewards. If the limitations of
AI and expectations have not been set out to staff, and if no one – other than the business itself - is held accountable for any misuse, issues will arise. The fundamental areas of potential risk are:

• AI can perpetuate bias and unlawful discrimination

AI is only trained on the information it’s given and, therefore, algorithms are only as unbiased as the data provided. Any biased data can then be amplified and inadvertently perpetuate bias or unlawful discrimination in decision-making.

For instance, if used as part of a recruitment process, the AI system is provided with information from a company that has historically hired men and therefore lacks diverse data, it may then ultimately ‘learn’ to prefer male candidates. Many AI systems are also lacking data in ethnicity and therefore, this in turn could result in discrimination against not only female candidates, contrary to the Equality
Act 2010, but also create unequal opportunities for people from diverse backgrounds.

• Employee privacy and data privacy

AI use will often involve the collection and processing of huge amounts of employee data and may contain sensitive or personal data. If mishandled, this could lead to employee privacy and data security breaches, or the unauthorised access or misuse of personal information. AI systems are also increasingly being used to monitor people, for example, through facial recognition systems. The Information Commissioner’s Office (ICO) has issued guidance and, in some instances, is banning their use due to concerns around the data risks.

• Unfair dismissal and breach of implied trust and confidence

Where AI tools are relied upon when determining an employee’s dismissal, there is the potential for irrational or unfair decision-making. It might not always be clear or transparent as to the reason for the determination, and managers may not understand how to interpret algorithms and data. It’s therefore important to retain some human involvement.

There is also an implied obligation of trust and confidence between employer and employee. Case law has established that employers need to provide clear reasons for important decisions and changes being made, and ensure that any such decisions are made lawfully, rationally and in good faith – even if the decision-making has been delegated to an AI tool. Where trust and confidence are breached, the employee may be entitled to resign and claim constructive unfair dismissal where they have two years’ qualifying service.

• Lack of empathy and human connection

It’s also important to appreciate that whilst AI may assist in many ways, it will lack human intuition, empathy and the ability to understand the nuances of human behaviours. Allowing AI to make all of the decisions relating to staff and their future careers can leave employees feeling disengaged, alienated and resentful. This, in turn, could result in a lack of job satisfaction and a workplace that is incapable of dealing effectively with conflict and staff issues.

There will always be a need for a ‘human solution’ in HR, to ensure that staff are given the appropriate level of support according to their individual needs, rather than a ‘one-size-fits all’ approach. Employee wellbeing and engagement should remain at the heart of HR.

Managing risk

To minimise potential risk and to achieve responsible and ethical use of AI, businesses should consider the following:

• Carry out an AI audit and risk assessment for using AI systems
• Train HR and managers to under-stand algorithms and check data accuracy
• Regularly audit and test AI systems and data inputs for bias
• Consult with staff and undertake pulse surveys
• Transparency regarding data use and collection and employee privacy
• Consider if reasonable adjustments are needed in respect of the use of AI for individuals with disabilities.
• Balance AI with the ‘human’ touch
• Ensure a ‘human’ manager has final responsibility for any decisions.

Policy and audit

Carrying out an AI Audit is an important first step to checking and establishing how AI is used in the business. Whilst many companies will have existing policies relating to social media and IT use, it would also be advisable to have a standalone AI policy to ensure ethical and responsible use.

An AI policy should outline what level of use is allowed, how improper use will be dealt with, signpost employees to where they can raise concerns and spell out the employer’s expectations and overall approach on employee AI use. An uncontrolled or unfettered use of AI could potentially expose a business to copyright issues and inadvertent sharing of proprietary company information. With the introduction of any new policy, it’s important to remember that clear and transparent communication and implementation is the key to its success.

In view of the potential risks involved in the misuse or unregulated use of AI, for which the business would ultimately be liable, it would appear logical for HR to be leading the way on internal AI use, just as it would for other internal practices and procedures. HR is able to address and review the potential risks of AI use in the workplace and can be central in releasing AI’s potential whilst navigating the company and its staff through these unchartered territories.

It’s paramount that HR effectively manages and harnesses AI usage in the workplace, rather than allowing AI to ‘hallucinate’ – leaving companies open to risk and possible claims.