At HumanFeedback.ai, we’re unlocking the final frontier of AI: truly understanding humans. Today’s models struggle with emotions, preferences, and intent because they rely on static, text-based data, and ignore how people actually feel and respond. Our platform solves this by delivering fast, reliable, time-based human feedback on video and other dynamic content. By combining advanced AI methods with insights from behavioral science, we filter noise, surface what matters, and deliver high-quality data with minimal effort, powering the next generation of human-centered AI across industries.
project page

Can out-of-the-box pretrained Large Language Models (LLMs) detect human affect successfully when observing a video? To address this question, for the first time, we evaluate comprehensively the capacity of popular LLMs to annotate and successfully predict continuous affect annotations of videos when prompted by a sequence of text and video frames in a multimodal fashion.
watch video

We surveyed the current state of the art in affective computing as applied to games (covering over 250 papers) and we provide a taxonomy of terms, methods and approaches used. We present the results along with a discussion on the current limitations of affective game computing and our vision for the most promising future research directions in the field! We hope that this project will act as a key driver of ground-breaking research and innovation in this emerging field!
watch video

The AGAIN dataset contains over 1000 annotated gameplay sessions, where players themselves annotate their playthrough in terms of the arousal they felt at each point of the game.
project page

Part of the Malta Energy & Water Agency's ‘Green Home Explained!’ educational toolkit, a comprehensive educational package created to help teachers integrate concepts on energy efficiency and renewable energy in their lessons in a fun and engaging manner.
play game
project page

In this project we built prediction models for viewers’ engagement based on data collected from the popular battle royale game PlayerUnknown’s Battlegrounds. We collected viewers’ chat logs and in-game telemetry data from several hundred matches of five popular streamers and machine learnt the mapping between gameplay and viewer chat frequency during play, using small neural networks.
watch video

The Platform for Audiovisual General-purpose ANnotation (PAGAN) was designed to address the issues of collecting large amounts of emotional data providing an easy-to-use framework for multi-purpose audiovisual annotation without installation, need for programming knowledge, or direct researcher supervision.
project page
project github

The Tensor-bAsed Machine learning towards genEral moDels of affect (TAMED) project led by Dr. Konstantinos Makantasis developed holistic affect-modelling methods by integrating physiological, behavioural, and in-game interaction data, focusing on relative emotional changes rather than absolute states. Partnering with major game developers, it uses player interactivity as a privacy-friendly, powerful source for personalising gaming experiences in real time.
project page
watch video

Originally developed for the Global Game Jam, Extreme Yoga Ultimate is a game designed around alternative controls. The game is based on Tobii Eyetracking technology and was designed to provide a hands-free experience. Extreme Yoga has been used to study gaze interaction in children’s gameplay.

Is it possible to predict the motivation of players just by observing their gameplay data? Our key findings suggest that gameplay features are strong predictors of player motivation as the best obtained models reach accuracies of near certainty, from 92% up to 94% on unseen players. The project was a collaboration between Ubisoft Massive Entertainment and the Institute of Digital Games (IDG), University of Malta.
watch video

The Preference Learning Toolbox (PLT) is an open source software application and package which supports the key data modelling phases incorporating various popular data pre-processing, feature selection and preference learning methods. The Python version of PLT was designed and developped primarily by Elizabeth Camillieri. PLT was developed at the Institute of Digital Games at the University of Malta with the support of Maltco Lotteries.
project page

MAZING was developed for projects focused on the Theory of Mind and Believability Assessment. The game implements a top-down model of Computer Frustration based on contemporary theory and emprical observation to simulate an increasingly frustrated agent behaviour.
watch video

The Summer School on Artificial Intelligence and Games offers an intensive learning experience for students and professionals interested in AI techniques and their applications in game development, combining lectures, hands-on exercises, and networking opportunities.
visit page

The Roleplay Research Retreat is an interdisciplinary gathering of scholars, designers, and practitioners exploring the study, design, and theory of role-playing games through collaborative workshops and academic discussion.
visit page

The 2025 Digital Games Research Association Conference (DiGRA2025) brings together international researchers to share cutting-edge work on games, play, and player experience across academic and industry contexts.
visit page

The Malta Global Game Jam is part of the worldwide game development hackathon, where participants in Malta collaborate intensively over a weekend to create innovative games from scratch.
visit page

OPtiMaL is a collaborative project between one of the leading centres of Al in games research worldwide - Institute of Digital Games, Malta - and one of the world's leading research groups in computational intelligence - SUSTech, China that aims to introduce generative AI algorithms that are capable of creating multimodal content with multi-dimensional diversity, which is aware of human behaviour and adapts to human experience.
visit page

Night Parade of One Hundred Demons is a musical game designed by Dr. Constatino Oliva, inspired by Japanese folklore, blending tactical play with narrative creativity.
visit page

IEEE Transactions on Games is a peer-reviewed journal publishing high-quality research on the development, study, and application of games, spanning computer science, AI, game design, and games user research studies.
visit page

Logo and website design for the 2022 Foundations of Digital Games conference, creating a distinctive visual identity and accessible online presence for the event.
visit page

Doors is a philosophical puzzle adventure game designed by Prof. Stefano Gualeni and Nele Van de Mosselaer that blends whimsical world-building with interactive storytelling to explore questions about perception and reality.
visit page

Construction Boom is a satirical tabletop game commenting on urban development and economic speculation through playful mechanics and exaggerated aesthetics designed by Prof. Stefano Gualeni and Jasper Schellekens.
visit page

Personal portfolio website of game designer and scholar Prof. Stefano Gualeni, showcasing his games, research, and creative projects.
visit page

The Tensor-bAsed Machine learning towards genEral moDels of affect (TAMED) project led by Dr. Konstantinos Makantasis developed holistic affect-modelling methods by integrating physiological, behavioural, and in-game interaction data, focusing on relative emotional changes rather than absolute states. Partnering with major game developers, it uses player interactivity as a privacy-friendly, powerful source for personalising gaming experiences in real time.

Logo and website for the LearnML project, an educational initiative teaching machine learning concepts through accessible interactive content.
visit page

Branding and online platform for the 2020 Foundations of Digital Games conference, delivering a cohesive digital presence for participants.
visit page

Conference logo designs for the 2019 and 2020 IEEE Conference on Games, visually representing the event’s focus on AI, game research, and interactive entertainment.
visit page

The Preference Learning Toolbox (PLT) is an open source software application and package which supports the key data modelling phases incorporating various popular data pre-processing, feature selection and preference learning methods. The Python version of PLT was designed and developped primarily by Elizabeth Camillieri. PLT was developed at the Institute of Digital Games at the University of Malta with the support of Maltco Lotteries.
project page

Website for the Experimental AI in Games (EXAG) workshop at FDG 2022, highlighting research on novel, playful, and unconventional uses of AI in games.
visit page

Website for the Procedural Content Generation Workshop at FDG, dedicated to research and discussion on AI-driven and algorithmic game content creation.
visit page

Website for the Tabletop Games Workshop at FDG, focusing on design, analysis, and player experience in analog games.
visit page

Website for the AAMAS 2018 tutorial bridging game theory and AI in games, offering a structured introduction to the mathematical and computational models behind strategic play.
visit page

The Newborn World is a storytelling game where players collaboratively shape the lore and myths of a newly created fantasy world designed by Prof. Antonios Liapis.
visit page

Website for Iconoscope, a social game of interpretation and creativity where players invent and guess meanings for visual symbols.
visit page