Does AI work in the workplace? (An article not generated by AI!)

The use of such technology is becoming increasingly popular, but does AI work in the workplace? Our employment experts look into why being cautious with AI may be the best form of approach.

Artificial intelligence (AI) has been around for many years, providing quick, 24-hour automated assistance in the manufacturing and distribution industries. It is now being increasingly used to help make HR decisions, as well as for recruitment, on-boarding and managing employees.

The automation of processing and analysing data in HR practices can lead to outcomes or decisions which are discriminatory and have a hidden bias, which – in the worst case scenarios – may lead to employment tribunal claims of:

• discrimination
• unfair dismissal based on a breach of mutual trust and confidence
• harassment and bullying
• breaching data protection/GDPR.

Recruitment

Candidates can use AI to produce a comprehensive CV that matches a job description in minutes. However, for the recruiter, the CV is a useful first impression of the candidate. An AI-produced CV may hide any potential shortfall in skills and may lack character.

Providing interview questions to a candidate before the interview may result in the use of AI-generated model answers that won’t demonstrate their ability to think on their feet. By not giving advance notice of the questions, the candidate can demonstrate their true ability, and the interviewer can ask additional questions that arise from the answers.

However, there may be instances when interview questions need to be provided in advance as a reasonable adjustment for an applicant who has a disability and/or is neurodiverse.

Additionally, recruiters may also use assessed tasks as part of the interview process, capturing the candidate’s skills and approach to role-specific issues which may provide a more reliable indicator of suitability.

Software is available to screen job applications for the use of AI. However, extreme caution must be exercised to prevent breaching data protection and GDPR rules. The recruiter’s privacy policy must reflect that these tools are used to screen applications, but also that there’s a right to review and challenge any decision made upon reliance of such tools. This will often require a data protection impact assessment.

One way to check if AI is being used by candidates is to include the company’s position on the use of AI in the recruitment policy/application pack. Requiring the honest disclosure of AI use during the recruitment process may also put off candidates from using it. It’s not discriminatory or unfair to ask a candidate in an interview if they have used AI, however their answer will need to be carefully considered before any outcome is determined to avoid any potential discriminatory issues.

During employment

Using automated tools during employment, redundancy processes, disciplinary or absence management procedures is also fraught with risk. Ultimately, any automated decision should be reviewed by a person to manually check for inadvertent biases or discriminatory results which may go against any of the ‘protected characteristics’ under the Equality Act 2010.

Employers should also have a policy on the use of AI by their employees, which – if breached – would amount to a breach of the employment contract. This helps to ensure that employees use AI within the company’s confines and in accordance with the company’s wishes, avoiding breaches regarding the confidentiality of clients/other employees.

This may be useful to limit liability for the company. If an employee uses AI in the wrong way and without checking the result before it goes to a client or to an external party, it could very easily lead to legal or criminal liability. A recent High Court case – where contempt proceedings were considered against a lawyer who had cited ‘false’ legal authorities in a witness statement – showed how failing to check the results of AI-generated work can lead to litigation. Employers should have an AI policy which safeguards against this kind of outcome.

A lot of people management is based on knowing and understanding those you manage, which is only something that can be built over years of working with them – it’s not something that can be replaced by AI. Applying a reasonable approach to everyone isn’t going to be easy if you’re intending to come to balanced and fair decisions and outcomes which are based on reasonableness and fairness. For example, AI cannot guarantee the sickness absence management of a colleague won’t be discriminatory if it’s screening the number of absences, but isn’t reviewing the reasons for them.

If a company does use AI in its employment processes, it carries the risk of action from the ICO if any automated decision adversely affects an individual and could breach GDPR. In those cases, the individual has a right to review the decision and to challenge it. Using AI without limit or effective controls can also increase a company’s exposure to litigation from all directions – not just employees or job applicants, but also customers, suppliers and third parties (for instance, if an employee has plagiarised content).

In summary, AI can work in the workplace – with caution! Without replacing the judgment of HR managers and employment lawyers, AI can be used to save time in simple automated tasks but must be used with caution. Giving the benefit of the doubt and behaving reasonably during the phases of the employment life cycle cannot be replaced by AI, either partly or completely.

If you and your business require expert guidance on the use of AI in the workplace together with data protection and GDPR, please get in touch – our team will be happy to help.

Our Legal 500-rated employment law & business immigration team are experts in guiding businesses of all sizes and backgrounds through a range of issues that may arise.