MyADB Recruitment Intelligence (MARI): Using Artificial Intelligence to Support the Recruitment Process
Implementation: 2019–2020
Technology type: Artificial intelligence (AI)
ITD’s Partner Department: Budget, People, and Management Systems Department (BPMSD)
Technology service provider: Impress.ai
In line with ADB’s Operational Approach: Promoting digital development and innovative technologies
ADB has a manual recruitment process. About a hundred applications are submitted for each open position, translating to an average of 45,000 submissions yearly. Recruiters spend several hours sorting, reviewing, and filing resumes.
Fatigue in reviewing could cause some applications to fall through the cracks. In addition, the process does not foster good line manager/recruiter experience. Furthermore, the lengthy hiring procedure and the slow response time can lead to ADB losing out on recruiting qualified candidates who lose interest or are hired by other organizations.
ADB tested whether the offered platform of Impress.ai could address these issues. The platform was customized to fit ADB’s requirements, with artificial intelligence used to automate some parts of the recruitment sourcing process. Candidates uploaded their resumes to the digital platform, which parsed the documents, detected gaps, and prompted users to provide missing information.
Natural language generation was used to provide candidates with a realistic pre-screening interview experience. The AI-based platform asked situational judgment questions to candidates tailored to ADB’s core values, as well as on position-specific skills and knowledge.
Meanwhile, machine learning enabled the platform to learn from its interactions with applicants, thus helping recruiters ensure that candidates were evaluated based on eligibility, competency, and technical proficiency. The platform also had a mechanism that allowed candidates to give feedback on their experience using MARI.
In effect, the platform performed the tasks of line managers and recruiters by evaluating candidates’ applications vis-à-vis the requirements of the position they were applying for. Weights were given to the available responses so applications were scored quickly. The system recommended the disqualification of ineligible candidates. Shortlisted candidates were notified by the platform, which also scheduled their interviews.
The platform, which was dubbed MyADB Recruitment Intelligence (MARI), was built in 2019. The platform was initially tested through an administrative assistant job opening in December 2019. Two additional tests—a national staff IT officer and a public management specialist— were conducted between February and March 2020 to train the bot to recognize the differences in job openings.
MARI showed that a digital assistant can screen applicants efficiently, saving time and money. Instead of manually evaluating applications, HR personnel were able to focus their time on reviewing the qualifications of shortlisted candidates and conducting interviews. They were also able to access the real-time updates and analytics on the dashboard, allowing them to identify bottlenecks in the hiring process to make the candidate workflow management smoother.
MARI improved candidates’ recruitment experience and reduced the drop-off rate from 47% to 21.5%. The candidates also gave positive feedback on MARI. The FAQs module had a 95.4% accuracy rate in its responses to candidates. However, MARI was de-prioritized due to larger human resources initiatives.