Hired by machine: How to pass the Turing test and get recruited
As AI systems are increasingly being used by HR to identify candidates for a wide range of job vacancies, can businesses use these AI tools effectively without bias or discrimination?
The use of AI across many business processes has continued to expand. HR is a business function that is vital for enterprises to get right. Locating candidates for each vacancy is time-consuming and complicated. What if HR could use an AI to help?
Recently, the House of Lords Liaison Committee published their latest ‘AI in the UK: No room for complacency’ report that concluded the UK has made good strides forward using AI in many contexts and scenarios, but warns better coordination oversight is needed.
The Committee also recommended creating an AI strategy using the Centre for Data Ethics and Innovation (CDEI) to produce national standards for AI development and deployment. The USA is already calling for full disclosure when an AI is deployed.
Speaking to Maddyness UK, Eric Sydell, EVP of innovation at Modern Hire — a HR platform that uses a range of technologies to support and define the hiring process — explained how AI is impacting HR in general:
“There are several ways that artificial intelligence can be used in recruiting talent. The mainstream adoption of AI for talent acquisition is concentrated in the early stages of hiring: sourcing, scheduling and screening.
“In these areas, AI can effectively reduce the administrative, manual tasks on recruiters’ plates, such as scheduling interviews and conducting phone screens, while also supporting a digital, enhanced candidate experience. The net result is faster hiring, and better candidate engagement since recruiters can focus more of their day on strategic recruiting work.”
Vincent Lonij is the CEO of Swyg, a new business that is reinventing how interviews are conducted.
He said: “Instead of using AI to judge candidates directly, we use AI to create profiles of interviewers to help us better understand their judgements and to help interviewers be fairer.
This gives us access to a broader pool of interviewers that is more representative of the candidate pool which, in turn, makes the entire interview process more inclusive.”
The drive to use more automated systems will accelerate. Using AI to help and support recruiters is rapidly expanding as businesses reduce costs, improve efficiency and streamline their business as they look to redefine their companies post-COVID-19.
Assessed by machine
Automated systems are increasingly being used in the recruitment process. According to a TUC report, the most commonly experienced technologies used by recruiters were automated CV ‘scraping’ — when CVs are scanned for keywords and a decision then taken as to whether a candidate will proceed to the next stage of the process (17%), automated background checking (16%) and video simulations or game-based assessment (14%).
Social media screening and AI-powered psychometric testing (both 11%) were experienced slightly less often by candidates.
The report concludes: “The use of AI in this way has significant implications for workers in terms of their employment rights, such as their rights to equality, privacy, and data protection, their physical and mental wellbeing, and broader issues such as the balance of power between employers and the workforce, and democracy at work.
“Despite this, many people do not know what these AI-powered management tools are, how they operate, and their impact. Indeed, it is very likely that these technologies are far more widespread than our survey results suggest.”
Modern Hire’s Eric Sydell advises: “Machines should not be permitted to make decisions based purely on applicant data. Instead, AI and machine learning should be used to support human decision making in the hiring process. By combing through application data and surfacing only the candidates that are the best fits for the role at hand, hiring algorithms can help humans make smarter hiring decisions — but at the end of the day, the machines aren’t responsible for any final decisions.”
How a business uses AI in its recruitment processes must be clearly defined to avoid unintentional bias and discrimination. As companies struggle under the increasing number of candidates applying for vacant positions, using some form of automation to identify potential candidates will inevitably increase. AI-based systems will be in the mix of technologies they deploy. Care, though, must be taken when setting up these systems to ensure they are fair.
The future of machine HR
No discussion of how AI is impacting recruitment is complete without closely considering the impacts of COVID-19. Businesses have had to transform their processes and workforces at speed. This includes the whole recruiting process.
Remote tools had become commonplace, but with the absence of any opportunities to meet, interview and assess candidates, remote and often automated systems have seen a massive rise in their deployment.
There is no doubt that an AI can analyse data more accurately than a human. Pre-assessment tools will increasingly have an AI component. What is vital is that the AI in use isn’t a black box. The AI’s outputs must be fully explainable, free from bias and discrimination.
These systems must have the confidence of HR professionals using these tools and demonstrate that candidates highlighted will be successful in the role they have applied for. How an AI is being used must always be disclosed to each candidate to form a trust foundation in the process.
“Recruitment is complex, and there is no single algorithm that is inherently ethical,” said Brian Mullins, CEO of Mind Foundry, which helps businesses and governments implement ethical AI systems. “The industry as a whole is generally focused on reducing the time and cost required to make a hire, and this means most people are not thinking about the systemic biases that AI can reinforce.
“It is easier to see the serious unintended consequences when a company uses AI to analyse candidates at the point of hiring, but more often than not, AI has already introduced bias into the process before even getting to that point.”
In 1950, Alan Turing developed what he called the imitation game — a test of a machine’s ability to pass as a human. Today, we are using intelligent machines to assess human skills and personality traits. These systems can be a great aid in certain contexts and scenarios, particularly where vast quantities of data need to be analysed and assessed. The key is to understand how these systems are setup and applied to a specific talent acquisition process.
By David Howell