SECTION 5
The ethics of AI

The ethics of AI

Despite acknowledging the benefits of AI, the complexity of integrating AI, data and TI into existing processes can be a barrier to unleashing its full potential.

Fear is a key factor. When it comes to personal information, quality, ownership and access to data is a significant concern, as well as a fear of robots replacing humans and unease about unconscious bias in algorithms. The speed at which new technology is developing can also leave companies torn between the risk of trying and the fear of lagging behind.

It's a combination of concerns over data, impact on jobs, the validity of the output and the sheer speed at which it’s developing that create the fear factor. The big question is how do we overcome the fear factor of AI? The fear of what it can and can’t do.

Jonathan Holdsworth
Head of TA, Rolls Royce

Are you afraid of using AI? If so, why?

I’m worried about getting it wrong
I don’t understand it
I’m not scared of
using AI

The use of AI tools is set to soar over the next 18 months. Approximately three out of four HR and TA professionals who haven’t yet used AI are planning to adopt it shortly. With 20% of HR and TA professionals admitting their AI is not officially integrated, it is critical that they build in time for research and rigorous testing with procurement teams from both HR and IT.

There are huge risks when it comes to AI – but companies shouldn’t be paralysed by it. As yet, there isn’t much regulation surrounding the use of AI, so companies must self-regulate and use common sense. The key to success is to fully integrate the right technology across the process and always be aware of the risks.

Human rights and AI

Right to privacy: be careful about AI data collecting, inferring about candidates and how using it. Check data is stored according to legal guidelines.

Right to equal treatment in recruitment: be aware of building AI algorithms on the right representative data samples. If data collected/sample data has inherent bias, then so will AI conclusions.

The rewards of AI in recruitment are clear - from assessment to onboarding, it can streamline talent acquisition by automating repetitive tasks, enhancing candidate matching, and ultimately saving time and improving the quality of hires. It can also be used to mitigate bias, ensuring fair and objective evaluations of candidates. Applicant tracking software, such as Pinpoint, for example, allows HR teams to anonymise candidates, hiding information such as gender and ethnicity.

Data-driven insights can also identify and address disparities within a business, and act as a force for good in terms of diversity, equity and inclusion approaches. AI models can play a crucial role in promoting diversity and equity at all stages of the employee journey, and AI recruiting tools can help companies balance the pool of candidates by improving the inclusivity of language used in job descriptions, and by sourcing candidates from underrepresented groups.

I think that we run the risk of undoing all of the DEI work because AI is only as good as the engineer who created it. You have to think of the demographic of those writing AI code. It is an advancement, but it is is taught by our society and what already exists on the internet. So when it gets ugly, it's our fault.

Heidi Stephens
Director of Talent Acquisition
and Professional Development, ABCD

87% of TA professionals are already using AI in assessment. If the decisions that AI makes are based on data that is affected by unconscious bias and/or a lack of complete or accurate data, it could mean disastrous results for organisations and candidates.

Companies need to make sure the data they're using to build and train their algorithms is representative. For example, if the majority of AI engineers are white and male, the data that they choose to include and the logic behind the algorithms may skew results, based on unconscious biases. AI is limited by original coding and unless it is trained too, will overlook equality and diversity. For example, if previously, most successful candidates went to a certain university or were a certain gender or age profile, (because of historical biases) AI results might lay a company open to exposure for discrimination.

Companies should regularly audit, test and train AI algorithms to ensure they are free from discriminatory elements, and secure and protect the data used. Companies should also have a trained human in the loop to monitor AI outputs.

Using AI can support better decision making and significantly reduce hiring bias, resulting in a naturally diverse workforce. The quality of AI depends on the strength of accurate data, however it shouldn’t be used for decisions in candidate selection.

Felix Mitchell
Founder and Co-CEO
at Instant Impact

Three AI principles

Companies have a responsibility to take AI seriously and build ethical principles for effective AI implementation. We believe in three key principles every company must consider to minimise bias when using AI:

01

Transparency

It is essential companies are transparent about when and how AI is being used. When a candidate is selected, a company must be able to explain its decision. AI built on deep learning AI models often struggle to return an understandable explanation for why decisions or recommendations have been made and can leave companies open to ethical and legal risks. For example, if AI has generated a shortlist of the first five candidates to review, it can negatively impact the other candidates in the process. Companies must ensure they are using algorithms that can clearly explain why decisions have been made to ensure candidates are treated equally and tell candidates how and where AI has been used.

02

Privacy

AI’s sophisticated ability to predict behaviour and preferences can violate our human right to privacy, particularly when it comes to using AI-driven technology for social media, such as programmatic advertising. For example, Cambridge Analytica’s use of unauthorised personal data to influence behaviour during the 2016 U.S presidential election and the Brexit referendum has led to increased scrutiny.

03

Fairness

To reduce human bias, companies need to make sure that the data used to build and refine AI algorithms must be representative. Historical data must be treated with caution. Diverse teams, testing and bias audits are all critical bias mitigation tools. Those creating the software must also be neutral and not allow their own bias to unintentionally creep in.

Case Study

AI legislation

In 2022, U.S. Equal Employment Opportunity Commission (EEOC), filed a lawsuit against iTutorGroup Inc, a tutoring company, claiming it used hiring software powered by AI to illegally weed out older job applicants. The EEOC alleged that iTutorGroup programmed online recruitment software to screen out women aged 55 or older and men who were 60 or older. It was the first case by the EEOC involving a company’s use of AI to make employment decisions.

For now, there are no explicit UK laws governing the use of AI and other algorithmic management tools at work. However, existing protections from discrimination under the Equality Act 2010 continue to apply to all forms of AI used in employment.

Other governments are starting to legislate on AI. The European Union is working on a new legal framework that aims to significantly bolster regulations on the development and use of artificial intelligence. The proposed Artificial Intelligence Act would classify AI systems by risk and mandate various development and use requirements. More than 150 executives of European companies expressed their concern in an open letter to the European Commission, and they believe the rules risk harming competitiveness yet fail to deal with potential challenges.

Bills addressing the use of AI in hiring have also been passed in US states including New York and California. For small HR and Talent Acquisition teams, keeping up with changing legislation across multiple countries or continents is unrealistic, and this is where the procurement team or recruitment specialists can help.

Governments around the world are trying to catch up legislatively with the rapid advances in AI, and some are moving faster than others.

Christine Morton
Principal Consultant,
Proxima

AI can complement HR and talent acquisition teams and bring huge benefits. However, this must be done ethically and transparently. And until legislation catches up with technology, SMEs should consider their own policies on individuals’ use of AI within the recruitment process.

²Reuters, ³House of Commons Library, ⁴ Financial Times