At the IBA’s recent Annual Employment and Diversity Law Conference, Founder and Managing Principal of People + Culture Strategies, Joydeep Hor was a guest speaker on the topic of Artificial Intelligence (“AI”) bias and data transparency in the legal workplace. Joydeep is Secretary of the IBA Diversity and Equality Law Committee, and he was joined by guest speakers from Latin America, Continental Europe and South Asia. Joydeep spoke to the difficulties that AI presents for managers and HR professionals in terms of the potential for inconsistent treatment, unequal opportunities and ultimately, unlawful discrimination, from the bias of AI in the workplace.

What is AI?

AI refers to digital computers performing tasks that would usually be performed by “intelligent beings” such as people. At the conference, Joydeep described how AI achieves this by being similar to the neuron connections in the human body. These connections are called synapses and form a complex network. It is through this network that the brain receives and processes information enabling the human to learn. He explained that certain branches of AI work with “artificial neural networks”. It is these programs that imitate the functioning of the human brain. Information is then entered as input on one side, processed, and the result is output on the other side.

 Advantages and Opportunities of AI in the workplace

All businesses, including law firms are investing in AI technology. The Wall Street Journal estimates that spending on AI by companies will grow from $8 billion in 2016 to $47 billion by 2020 – which is an almost 600% increase. Joydeep attributes this increase to AI’s ability to optimise business processes by enhancing efficiency and automating time consuming administrative tasks. For example, in a legal context, AI is able to review contracts and undertake legal research almost instantly. One such software is “ROSS Intelligence”, which was used in a recent test where a partner at a law firm was able to find a matter that was identical to the firm’s case almost instantly, while his own research took ten hours. Joydeep and the panel noted that the research was not only efficient, but highly accurate.

However, at the conference, Joydeep questioned whether productivity control is automatically accompanied by greater effectiveness at the workplace. In making this point, Joydeep gave the example of the Henn Na Hotel, a hotel in Japan almost entirely staffed by robots. The hotel, shortly after opening, had to dump half of the robots due to them “annoying” the guests, not being capable of the jobs they were designed for, or creating more work for their human counterparts.

Risk of AI in the Workplace

Joydeep also raised the impact that AI has on the hiring, firing and promotion decisions of HR professionals. AI has been introduced recently to screen CVs to find the best possible applicant. Joydeep therefore raised how the lack of transparency in recruitment decisions may cause a problem for employers. In illustrating this point Joydeep highlighted that the vast increase in data sources may result in employers not being aware of what information has been used to make recruitment decisions and therefore what risks may arise. This problem is compounded by the fact employers and work councils often do not understand the numerical codes according to which the algorithm acts. This may mean employers cannot guarantee that an AI system has not made a recruitment decision that does not carry a risk of a discrimination claim. Joydeep recommends that in order to prevent these claims from arising employers should monitor the data input at the very initial levels and regularly review the systems they use. He suggests that employers using AI should prepare themselves to explain why one applicant was hired and another one rejected.