Part Three: HR AI. Ethics and Trust
Updated: Dec 8, 2022
How well do you know your HR technology stack? Is it trustworthy? Ethical? For the final part of this series, we'll turn to our role as HR in this context. In the face of near future laws and audit guidelines, what responsibility do we have as HR to help safeguard the use of AI for our employees and candidates?
Let's get the dry but necessary info out of the way first: the legal and compliance around using AI tools in HR is rapidly changing. Although not yet mandated, there are key government policies and frameworks that we should be watching closely for implications for future tools we use and considering our existing HR systems. Let's take a look at a few examples of what might be coming our way:
The state of Illinios laws now requires candidate consent for using video analysis.
Maryland state law restricts the use of facial analysis via technology for hiring purposes.
New York City Bias proposal aims to upgrade existing laws around anti-discrimination in hiring to fit a world that involves AI in hiring decisions. This includes telling candidates that tech has played a role in assessing them and vendors would also be responsible for annual audits to remove discrimination. If passed, this will take effect from January 2022.
European Union AI Regulation on "artificial intelligence systems".
This proposal made in April 2021 could have implications for the vendors who provide the technology and the companies who use the platforms (i.e. us as HR!). Some focus areas are not too different to those in GDPR, such as fairness, transparency and individual rights to access and control their data.
So what is new? There are many design, compliance and audit criteria listed in the proposal - read here from hrexecutive.com's John Sumser for an excellent summary. There is one significant area that impacts us all - both vendors and users of high risk AI:
Understand the output of the AI decision and use it appropriately and fairly. In other words, don't just blindly accept the decision made by an AI system without a human scrutinising both how the decision was made, and if the prediction or ranking is fair.
A recent article by the Society for Human Resource Management (SHRM) interviewing legal experts summarised it succinctly as:
"There are ethical implications of AI having an increasingly greater impact on individuals' lives,". "The EU proposal recognises that human intelligence cannot give way entirely to artificial intelligence and that there are boundaries into which it should not stray."
Amy Bird - Attorney, Clifford Chance, London
In the Australian context, we currently don't have laws specific to HR decisions made using AI, instead of relying on other existing legal frameworks around privacy and data security to provide guidance in this space. However, the Australian Human Rights Commission has completed a 2020 study Using artificial intelligence to make decisions: Addressing the problem of algorithmic bias, which is the start of a movement towards more formalised legislation in this area.
We're at a precarious moment in HR with our use of AI making or assisting in HR-related decisions. Although not yet legal requirements, the implications of the US and EU based rulings could have impacts that global and local HR industries need to address quickly.
Until the above areas become a legal requirement, here are a few ideas we should/could be doing as HR. Not only get the value from the AI HR tools, but also help safeguard our employees and candidates.
Build Trust in HR tools through validity, reliability, and transparency
As an individual HR practitioner, ask yourself the following questions when next using a HR tool to make a hiring or performance decision:
Validity and reliability
Do you understand what the AI is assessing for?
Does the AI tool measure what it claims to, and is it consistent?
Is the assessment relevant to the job for which a person will be doing?
Does it measure the specific job characteristics?
Does it remove discrimination and reduce bias?
The ethical lens through transparency
Is it transparent to candidates and employees that AI has been used?
Is it transparent to the candidates or employees what data has been collected, and their choices to opt-out?
If you don't know the answers, should you be using the tool? Seek out people in your organisation who can help you learn.
Design, Compliance and Audit
(sounds like a fun meeting to be in doesn't it! 😊)
Coming up a level up from the individual actions to the org level i.e. those of you responsible for product management of AI HR tools. There are some great resources to get started:
For Humanity: is a non-profit with the goal of "mitigating risk from the perspective of Ethics, Bias, Privacy, Trust, and Cybersecurity in our autonomous systems will lead to a better world". The community is based on crowdsourced input by industry leaders and not only is everyone welcome, but all perspectives are encouraged. They have several tangible frameworks that can be leveraged now. An excellent start is their article on removing bias in datasets that are used to train and test the AI algorithm.
Centre for AI and Digital Policy (CAIDP) is a non-profit that brings together experts across law, technology and organisational leaders to promote existing and established AI policy and frameworks including training via policy certification.
Engage with your internal legal and compliance teams + your vendors to work together.
Videos and podcasts suggestions:
Human AND Machine
Our future looks to be a hybrid of taking AI HR tools but teaming up with the human, ultimately deciding, albeit more accurate and bias-free - if we can get the mix right.
This man-machine hybrid is not a fantasy, it is backed by a Journal of Applied Psychology, meta-analysis ( a fancy way of describing a study that combines data from lots of other studies - a gold standard in scientific research). This study showed that when AI was used as a supporting tool for a human decision, it increased the accuracy of predicting future job performance.
Ultimately we should be aiming for an outcome where introducing AI into HR hiring decisions means positive outcomes for all - reduced cost and greater accuracy for organisations, whilst delivering a more transparent and fair process for candidates and employees.
About Talent Tech Solutions
Talent Tech Solutions advise and provide solutions to ensure our clients win with Talent Tech. We understand that our clients’ reputations are built on the successful delivery of Tech projects. We’re here to help our clients achieve project objectives and realise ROI without the pain. We do this by guiding clients at each and any stage of the project: selection, implementation and ongoing optimisation.
With over a decade of local and international experience we add capability and capacity to our clients existing team and leverage project management, consulting and implementation best practices.
We're technology agnostic and specialise in Applicant Tracking Systems (ATS), Candidate Relationship Management (CRM), psychometric testing, video interviewing, reference & background checking technology and system integrations.