In this project, we aim to make use of generative AI Chatbot technology to replace human actors, facilitating police trainees to interact with an AI suspect and practice their questioning and cautioning skills.
A vertical touch screen monitor displaying an AI avatar would form the Chatbot kiosk. The look of the AI avatar will be different based on the scenario designed. There will be a number of different scenarios to simulate different types of incidents/offences. Trainees will take turns to receive this training individually at the kiosk. Random scenarios would be introduced to the trainee in text and voice-over. The trainee is required to ask the AI avatar questions seeking to establish the key elements of a particular offence committed by the AI avatar. Once the trainee determines from the AI avatar’s replies that an offence has been committed, the trainee will have to cite the correct caution. The kiosk will verify whether the offence cited and the wordings of the caution are accurate, and if incorrect, it will provide the correct answer and offer recommendations on the trainee’s performance to complete the training.
All scenarios would be prepared by instructors in text format in advance. The information should include the scene (e.g. inside a supermarket), the suspect (e.g. age, gender, outfit), the case background (e.g. a suspected shoplifting case in which a particular item has been stolen), the offence committed (e.g. theft), the key words that constitute the elements of that offence and the correct wording of caution. Based on the text information, the AI technologies will generate the appropriate background image of the venue, the appropriate AI avatar, and the appropriate responses to the trainee’s questions.
