AI recruitment software provider Sapia.ai is set to showcase innovative research across four sessions at the upcoming SIOP 2023 Conference in Boston.
With the largest submission from an Australian-founded AI company, Sapia is proud to present cutting-edge research that will greatly catalyze the evolution and adoption of ethical AI, Natural Language Processing, and Machine Learning in the field of industrial-organizational psychology.
Sapia.ai has four individual research reports that will be presented at the conference, each of which is set to make a significant impact on the field of AI:
- Leveling the Playfield: A Collection of Inclusive Approaches to Assessment. This session examines the findings of Sapia’s research paper, Reducing Disability BIas with Chat-Based Interviews. It contends that in-person and traditional interviews are stressful and anxiety-inducing for candidates with disabilities, contributing to lower rates of representation and higher rates of unemployment. The session examines how an ethical AI, chat-based assessment can result in fairer outcomes for people with disabilities: a candidate happiness score of 8.9/10 and 95.8% positive or neutral sentiment when using Sapia’s AI Smart Chat Interviewer.
- Natural Language Processing (NLP) Applications: The Versatility of Text-Based Data. Showcasing findings from the paper Identifying and Mitigating Ethnicity Bias in Structured Interview Responses, this session explores how racial bias and racism in traditional job interviewing can be mitigated with the use of a blind, text-based interview format that uses underlying feature scores (i.e., personality traits, competencies, and communication skills) rather than raw text in the final scoring algorithm.
- Machine Learning for I-O 5.0: Addressing Algorithmic Bias and Fairness Concerns. This session explores the findings from the paper Plagiarism Detection and Prevalence in Online Text-based Structured Interviews. It will explain how Sapia’s AI system can flag plagiarized or faked responses in text-based chat interviews, mitigating the issue of ‘faking’ present both in AI-based and traditional self-report assessments. The results to be presented include Sapia’s own plagiarism rate of 3.28%, which is significantly lower than the 30-50% of candidates estimated to fake self-report measures.
- Machine Learning for I-O 5.0: Advances in Organizational Applications. Referencing the paper Question-aware Outlier Answer Detection for Fairer AI Scoring of Interviews, this session explores how an ethical AI can identify outlier responses that go against training data, leading to better outcomes for interviewees who respond to prompts in novel, but relevant, ways.
Sapia.ai’s participation in the SIOP 2023 Conference highlights the company’s ongoing commitment to discourse within the SIOP and wider I-O communities.
Dr Ashlie Plants Ph.D., Lead People Scientist, said that she was proud to see so much of Sapia’s innovative research accepted by the SIOP committee.
“Even in psychological circles, AI has been met with a great deal of apprehension and scrutiny, which is absolutely necessary,” Plants said. “But investigation and curiosity is also necessary, and we’re thrilled to see the response that the I-O psychology community has had to the progress and potential of a truly ethical AI system.”
“We can’t wait to present our research at the SIOP Conference, and to continue our hard work in powering Sapia’s AI Smart Chat Interviewer, ensuring it is the fairest and most valid means of interviewing and assessing candidates in the unfolding AI era.”
2023’s SIOP Conference will mark the third appearance of the Australian-founded company. The North American arm of the team will also be present at the conference.