Interviews are widely used for selecting personnel with varying standards, ranging from casual conversations to structured questions with standardized scoring systems. However, interviews cannot be conducted with all applicants because of high resource demands; they are typically used for a short list of candidates. Artificial intelligence (AI) technology now enables interviews to be performed from the early stages of selection.

AI interviews involve applicants answering questions on a computer screen, which may consist of one or multiple questions. The program records the video and audio of the responses, converting speech into text. Additionally, tone, pauses, clarity, and facial expressions are recorded. These factors can predict desired internal qualities, such as personality, leadership, creativity, or other company-specific variables, like job performance or turnover tendencies.

AI interviews are popular in the United States, with some consulting firms claiming to have conducted over a million AI interviews in 2019. This method is favored for its cost-effectiveness and the belief that it can provide accurate insights into an individual's qualities. However, evidence supporting the effectiveness of AI interviews still needs to be made clear. For instance, if measurements are taken twice with a specific time interval (also known as test-retest reliability), the correlation between them is low. AI personality assessments may not align with those assessed by interviewees (through self-reported personality tests) but rather with the interviewer's personality ratings. This raises questions about whether AI interview scores reflect an applicant's personality or indicate the direction of their answers without capturing their true essence.

Although evidence raises concerns among scholars, developers learning to create effective questions and more data for training interview models have led researchers to hope that AI interviews can yield good results. AI training technology has also advanced, from counting words in language analysis to using language models (i.e., transformers) like BERT and GPT. This, combined with improved voice and facial expression detection, results in more data extracted from applicant video clips. With these technological advancements, AI interviews are expected to gain popularity worldwide.

Practical applications. Human resources professionals should start recording employee interviews and strive to capture clear audio and facial expressions. Interviews should have a structured format with standard questions for all interviewees. This data will be valuable in training AI models in the future. Proper data collection will result in more accurate AI models. For those being interviewed by AI, practice answering questions within a limited time. Training will help you avoid hesitation and make your responses more relaxed and structured, resulting in a more accurate AI assessment of your potential without penalizing you.
Our services. We are currently experimenting with developing AI interviewer in Thai. If organization leaders or HR officers have an interest in investing in the AI service, please do not hesitate to contact us through our website or Facebook direct message.