4 pillars for using AI responsibly in a skill-based organization
By Jeroen Van Hautte, Co-Founder and CTO and Andreas De Neve, Co-Founder and CEO, TechWolf
- As the world moves towards skill-based workforces more companies are drawing on AI to support them with this.
- This makes it crucial for AI to be used responsibly: without clear oversight, systemic biases will worsen in workforces and employee trust will be lost.
- By following four key pillars for using AI responsibly, skill-based organizations can ensure a transparent approach that benefits employees and employers.
Jobs are changing too quickly for traditional workforce models to keep up. This has given rise to a new term, the ‘skill-based organization.’ Studies indicate that 90% of business executives are now experimenting with building a skill-based organization.
Moving from jobs to skills is complex. It involves unravelling roles to a granular, skill-based level. That takes time and there’s potentially missing data if an HR or learning system hasn’t been updated in a while. Plus, the number of skills that an organization has is approximately twice the number of roles and jobs.
Organizations have realized that keeping track of dozens of continuously evolving skills across thousands of workers is no feasible manual task. So they turn to AI. This is a step in the right direction and the only way we can truly embrace the potential of the skill-based organization. But, we must tread carefully.
When people’s livelihoods are on the line, the AI used to influence those lives must be as open, transparent and ethical as possible. Tales of AI-gone-bad are, unfortunately, all too common. Amazon had to cancel a recruitment AI model due to bias against women. More recently, concerns have been raised that hiring and screening algorithms discriminate against people with disabilities.
Key pillars to using AI responsibly
It’s vital that AI is used for good in the skill-based organization, or all of the effort put into shifting away from jobs will be for nothing. Forming the basis of this approach are four key pillars for using AI responsibly:
Pillar 1: Know your data sources
An AI tool is only as good as the data sources it works with. Give it inaccurate data and you’ll get inaccurate results. AI, working at scale in your organization, can drastically increase bias if it is modelled on biased or incomplete data. Before you start giving your AI skills data to train on, you must audit your data to ensure it’s as accurate and representative as possible.
Ultimately, not all of your data will be objective or equitable. For example, in our work at TechWolf, we’ve discovered that male employees tend to over-report their skills, whereas female employees tend to under-report them. If you fail to understand such contextual factors when using skills data, you’ll find that your system excludes important talent. Undermining the entire point of the skill-based organization.
When evaluating data sources, organizations should keep four properties in mind:
A final note on this, we mandate using non-invasive data sources to avoid infringing on employee privacy. Avoid using invasive data sources such as email, private chat or any other data sources that could be seen as prying. A good check for this is to ask yourself if you would be comfortable if your manager had access to the data sources you’re intending to use — your emails, your Slack messages and so forth.
Pillar 2: AI’s decisions are explainable
Robert Jones went down in history when he drove off of a cliff edge while blindly following his sat-nav’s instructions. When acting on an AI’s recommendations, you must understand how it has come to its conclusion. Algorithms that can explain their reasoning are more trusted and suffer from less bias.
That’s why there’s a host of legislation aimed at enforcing a certain level of explainability. In the EU, GDPR provides employees with the “right to an explanation” when algorithms measure or evaluate aspects related to them based on automated data processing. The AI Act is also making its way through the European Parliament. In the US, the National Institute of Standards and Technology has published its Four Principles of Explainable Artificial Intelligence.
Pillar 3: Employees own their data
Employees generate data throughout the work day, with every document written, project finished and course completed. Ideally, the data generated and skills inferred from this will be portable across organizations and used by individuals to grow their careers. But at the very least, individuals must have the final say on whether skills are part of their profile or not — and if AI can use it.
Key to this is highlighting (and delivering on) the benefits of using skills for workforce decisions. Taking this approach suits employees, with 79% stating that they’d be okay with having their skills data collected by employers and a further 14% open to it, depending on the purpose.
Pillar 4: Share the uses and benefits
Employees are open to sharing their skills data with their employers, but many only wish to do so if the benefits to them are clear. If they get fairer hiring, tailored work experiences and growth opportunities, they are happy to share data. Additionally, let employees know how their data is collected, from where and how the AI uses it. You might need to do a bit of upskilling around AI to ensure all employees understand these points.
The four pillars: holding up the skill-based organization
The world is moving towards a skill-based approach and that increases AI’s role in workforce decision-making. It’s not a matter of if, but when AI comes to the skill-based organization. Set the right foundations by integrating the four pillars into your AI strategy. Without them, everything will come crumbling down.
This article first appeared on the World Economic Forum blog.
Header image credit: Karolina Grabowska