Sapia Labs, the research and development arm of Sapia.ai, is set to present four new research papers on the impacts of AI and chat interview technology on hiring processes at the forthcoming Society for Industrial-Organisational Psychology (SIOP) conference. The event, due to be held in Chicago, Illinois, on April 17, will be a fitting stage for the company's pioneering work that seeks to augment global understanding of cutting-edge AI technology.
The research resonates across several key areas of the AI domain, ranging from analysis of the personality traits of ChatGPT, to the competent employment of large language models as a tool for scoring and elucidating structured chat interview results. By delving into the facets of personality projection by different AI models, the research promises to enhance the effectiveness and reliability of AI in recruitment.
Buddhi Jayatilleke, the Chief Data Scientist and Head of Sapia.ai Labs, vocalised the existential urgency for businesses in comprehending and harnessing the power of AI. “Every business is thinking of how to adopt AI. This work aims to contribute to the broader understanding of AI as well as its effective adoption in the hiring process,” Jayatilleke stated.
Underpinned by Sapia's proprietary database, consisting of over 15 million responses from 3.5 million global applicants, Sapia Labs independently strategised its research, thus corroborating certain features of AI technology. “We aim for our conclusions to be both tangible and practical, as to add to the academic debate around the use of AI in various contexts,” Jayatilleke added.
Empowering numerous high-profile brands such as Qantas, Woolworths, and Starbucks, Sapia.ai utilises AI-driven conversational interfaces to assist the recruitment process. Their innovative process assembles insightful data from candidates, promotes candidate screening devoid of bias, and aims to foster diversity in recruitment.
The newly tabled research papers offer fresh perspectives on AI employment in hiring practices. One study, evaluating candidates’ reactions to different types of online interviews, revealed a conspicuous preference for chat-based interviews over video format, particularly among women and several racial minority groups.
Another substantial contribution is an upper hand in grading and explaining interview responses using large language models. The study offers a novel avenue for dependable and scalable interview scoring. An investigation into mapping between Sapia.ai's Communication Score and the Common European Framework of Reference for Languages (CEFR) levels beefs up the validity, reliability, and fairness of the scoring system.
Finally, a paper assessing the personality traits projected by different AI models, establishes GPT4, with its human feedback loop, as having higher scores on honesty, humility, agreeableness, extraversion, and consciousness when contrasted against GPT-2 and the global population benchmark.
This new work by Sapia Labs not only serves to codify the capabilities of AI in hiring, but also enriches academic debates and insights around AI applications. With this, one can expect a brighter and more inclusive future for hiring practices globally, the company states.