Periodic Reporting for period 1 - TAWNY (TAWNY - An Artificial Intelligence Project to Make Things Empathic)
Reporting period: 2018-05-01 to 2018-09-30
TAWNY is an AI startup with the goal to develop an Emotion AI that allows the world’s machines and digital products to be empathic, i.e. to seamlessly and dynamically adapt to a user’s needs and preferences at the very moment. In contrast to existing approaches in this area, TAWNY is not limited to basic emotions like “being happy” or “being sad” but aims to develop an Emotion AI that has a holistic understanding of what a human is, allowing it to adapt to recognize specific mental states like being in a state called “flow” – the state of optimal experience, e.g. being very productive without feeling stressed. It was the goal of the feasibility assessment to show that such a flow recognition system is technically feasible and economically viable. In the technical feasibility study, we showed that using our Emotion AI it is possible to automatically measure flow using physiological signals from wrist-worn devices. In order to do so, we calibrated and fine-tuned our Emotion AI technology towards this new task of flow detection. The study setup was based on participants playing a computer game, inducing different states like boredom, flow and stress. To examine the economic viability of our technology and to refine our business plan, a market research study was conducted. An online survey was carried out involving 15 decision makers from different European companies from various industry sectors. The market research showed that our general idea is economically viable and the way we want to approach the market makes sense. Consequently, our project shall continue and we want to scale our Emotion AI system in the next steps.
Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far
For the technical feasibility assessment, we followed the four-steps approach that was outlined in our proposal: i) Presence: Can physiological signals be reliably measured with wearables during the activity of playing a computer game? ii) Validation: Can physiological signals in this scenario be correlated to an actual mental state like boredom or flow? iii) Classification: Can we automatically classify physiological signals into different states? iv) Action: Can we use such a flow classification system to increase the performance of the players? By re-training and fine-tuning our Emotion AI, we could confirm all these questions, proving technological feasibility of the approach. To assess the economic viability of our approach, we conducted a market research with regard to AI in general and specifically Emotion AI. The market research was executed in the form of an online questionnaire. We surveyed 15 companies of different industries. We specifically asked decision makers about the willingness to pay for AI solutions, possible product packages and possible payment models (e.g. pay as you go, consulting, etc.). There were specific questions about Affective Computing and Emotion AI, which should also stimulate to think about emotion and affect detection technologies like our Emotion AI and its integration in the daily business. Overall the survey showed, that there’s a high interest in Emotion AI technologies just like as in AI technologies in general. (Top-level) managers of large European companies believe in the large potential of AI and see AI-based innovation as a strategically critical factor for Europe to stay competitive with China and the US. Furthermore, TAWNY’s current mode of operation, i.e. providing off-the-shelf Emotion AI optionally tailored to client-specific use cases, fits well with the current way large companies adopt and apply AI (i.e. buying off-the-shelf solutions and custom external development). The flow classification system developed in this feasibility assessment is directly related to the current main drivers for AI adoption, i.e. efficiency, cost-saving, and enabling completely new products. Consequently, we want to take next steps towards further commercialization of our solution. We also want to present the work of this project at tech conferences, trade fairs, etc. to raise awareness for out technology.
Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)
Using TAWNY’s Emotion AI technology, it was possible to create the first-of-its-kind machine learning model which allows to distinguish between the three states boredom, flow and stress more than 50% more accurately than with a baseline approach. As we have learned from our market research, companies are very much interested in such a system. The proven technological feasibility signifies another step towards a new era of human-machine interaction with the goal to improve and facilitate people’s lives.