12 Information sheet for participants Affective Human-Chatbot Interactions Based on Multimodal User Data Participant (full name): Conducting person (full name): Time and place: June 2022 July 2022, online Contact project team: Nikola Kovacevic, Study Coordinator, nikola.kovacevic@inf.ethz.ch Data Protection Officer ETH Zurich: Tomislav Mitar, tomislav.mitar@sl.ethz.ch Please read the text below carefully and ask the conducting person about anything you do not understand or would like to know. Information about the study What is the purpose of this study? The purpose of this study is to collect a dataset that will serve as a basis for several projects investigating human-chatbot interaction. Concretely, this study should help the development of more intelligent and engaging chatbots for various applications such as health care, education, and entertainment. Our goal is to equip chatbots with a formal understanding of emotions and personality, allowing the chatbots to both infer emotions from the user and having an emotion and a personality itself. This would make the conversation more human-like and engaging for the users. What is investigated and why? We will investigate how different chatbot responses influence human emotions and how humans perceive chatbot behavior by letting participants converse with three different chatbots (stylized Albert Einstein and a generic male and female character) through speech on a custom data collection webpage for 30 days . You will be asked to sporadically interact with the chatbots for a few minutes on at least ten different days and periodically submit self-reports about your own emotional state and how you personality. The voice and the camera input will be recorded while speaking with the chatbots. From the collected data, we will develop emotion-aware chatbots, which can find various applications in education, health care, and entertainment by adapting to user needs and thereby improve student participation, mental health therapy, and user experience, respectively. Who can participate? Participation is only possible if all the following criteria are met: You are between 18 and 60 years old. You are regularly using a computer that is equipped with a front camera, a microphone and speakers (or a headset), one of the following four browsers Chrome, Safari, Edge, Firefox, and access to the internet. You can solidly communicate in English. You do not take any kind of drugs (including Marihuana). 13 You do not take tranquilizers or psychotropic drugs (anti-depressants) and are not affected by any type of the autism spectrum disorders (e.g., Asperger). You have given your consent. Will I be compensated for participating? You will be compensated with CHF 60 (base compensation) if your participation meets the following requirements: 1. You fill in both the entry and exit questionnaires. 2. You fill in at least four self-reports per day on average, which corresponds to approximately two hours of active participation over the entire study. 3. You were active on at least ten different days (i.e., one or more self-reports on ten different days) You will be additionally compensated with CHF 50 (extra compensation) , if your participation meets the following requirements: 4. You fill in at least eight self-reports per day on average, which corresponds to approximately four hours of active participation over the entire study. The maximum compensation that can be reached is CHF 110 . You will be able to track your progress on the webpage and compare yourself to other participants through an anonymized leaderboard based on the average number of self-reports per day. The interval after which self-reports become available depends on your engagement. A self- report becomes available approximately every 60 to 90 seconds of active conversation with the chatbot if the following criteria are met: 1. The camera is not occluded, and your face is visible most of the time (no face occlusion or undetected face for longer than three seconds). 2. The intervals between received and sent messages is shorter than 30 seconds (i.e., you respond after at most 30 seconds). 3. The content of your message contains meaningful text (i.e., is not gibberish). There will be a lottery at the end of the study where one participant will win CHF 00 . Each participant that concludes the study and fills in the entry and exit questionnaires will receive one lottery ticket and automatically participate in the lottery. Additionally, the participants can increase their chance of winning by collecting more lottery tickets in the following way: A total of 45 lottery tickets will be distributed among all participants based on the final leaderboard (i.e., based on the average number of self-reports per day): 1. place: 20 lottery tickets 2. place: 15 lottery tickets 3. place: 10 lottery tickets 14 You can receive up to 20 additional lottery tickets by unlocking the following badges: Bronze badge: 30 self-reports 1 lottery ticket Silver badge: 100 self-reports 5 lottery tickets Gold badge: 250 self-reports 10 lottery tickets Platinum badge: 400 self-reports 20 lottery tickets The winner of the lottery will be determined two weeks after the study finishes. We reserve the right to exclude participants from the study in case of obvious cheating or misbehavior. Excluded participants will not be compensated nor will they be able to participate in the final lottery. How much time will I be expected to invest? You will be expected to invest at least two full hours in total of actively conversing with the chatbots throughout the course of the study spread over at least ten different days (see base compensation). Additionally, you will be asked to fill in an entry and exit questionnaire, which will take about 10 minutes each. For reaching the maximal compensation of CHF 110, you will be expected to additionally invest two full hours of conversation with the chatbots (see extra compensation). Are there any risks or benefits? There are no imminent risks to participating in this study. However, the chatbot responses are passed through from a third-party chatbot provider and are not preprocessed nor censored or manipulated in any way. We can therefore not avoid insults or inappropriate messages from the chatbots. However, we emphasize that the chatbots do not understand context nor are they messages are therefore not to be taken personally but are to be considered as fully fictional. All data will be stored securely on servers of ETH Zurich and will not be distributed or shared with individuals or third parties other than the involved research groups. Furthermore, your data will not be stored together with your personal information, but separately and anonymized through a random identifier, guaranteeing full anonymity. How am I insured? Adverse health effects that are directly caused by the study and can be demonstrated to be attributable to fault on the part of the project team or ETH Zurich are covered by ETH's liability insurance (Basler Versicherungen, policy no. 30/4.078.362). You are responsible for insuring yourself against any other adverse health effects that might occur. What are my rights during participation? Your participation in this study is voluntary. You may cancel your participation at any time without specifying reasons by writing an e-mail to the study coordinator. However, you will not be compensated if you withdraw before the end of the study, and you will not be eligible for the lottery. Which data are collected? We will collect all voice messages that are exchanged with the chatbot during the study and the camera recordings that overlap with voice input. Additionally, you will be asked to fill in an entry and an exit questionnaire containing demographics and a personality test. Furthermore, we will ask you to fill in several self-reports assessing your current emotional state during the conversations with the chatbots. Additionally, you will be asked to give subjective estimates of the chatbot 15 be anonymously stored on servers of ETH Zurich. Any future publications that result from this study will exclusively be carried out using anonymized data. How are my data treated? Your data will be stored securely on servers of ETH Zurich. The data will be anonymized through random identifiers and stored separately from your personal information. The recorded voice snippets will be translated to written text through Google. The resulting written text will be sent in chunks to two third-party chatbot providers for generating chatbot responses. The video recordings will not be provided to any third parties. All data will be stored securely on servers of ETH Zurich. The data will further be used for the development and validation of emotion-aware chatbot models. However, at no time will your identity be inferable from the data. Members of the ETH Zurich Ethics Commission may access the original data for examination purposes. Strict confidentiality will be observed at any time. What are my rights to the data? You may request comprehensive information about the personal data that were collected from you in the study at any time. You also have the right to have them corrected, handed over to you, barred for processing, or deleted. You may revoke your consent to the processing of your personal data at any time without giving reasons. Your contact person for related requests is given above. Who funds the study? The study is funded by the Computer Graphics Lab of ETH Zurich. Who examined the study? This study was approved by the ETH Zurich Ethics Commission as proposal 20XX-N-XX