Home
Research
Ethics and Human Values in Technology
ALFIE: Enhancing Ethical AI Through Dialogue and Innovation

Project

Funding agency: HORIZON Innovation Actions, European Commission-EU

Duration: 1/10/2024 - 30/9/2027

Funding agency: HORIZON Innovation Actions, European Commission-EU

Project type: Innovation project

Coordinator: CATALINK LIMITED

Principal investigator: Juraj Podroužek, Martin Tamajka

Project web: https://alfie-project.eu/

ALFIE: Enhancing Ethical AI Through Dialogue and Innovation

The ALFIE project combines an EthiTech Dialogue Hub (ETD-Hub) with an advanced AutoML platform and interaction in natural language to connect ethical discourse and practical AI implementation. This initiative aims to contribute to the development of responsible AI practices across Europe by fostering inclusive discussions and translating ethical insights into AI solutions.

ALFIE project consists of two main components: the EthiTech Dialogue Hub (ETD-Hub) and the AutoML platform enhanced by interaction in natural language. The ETD-Hub serves as a multi-disciplinary forum where citizens, policymakers, AI professionals, and legal entities engage in conversations about the legal and ethical aspects of AI technologies. These discussions inform policy recommendations that may influence the European legal framework on AI, helping it adapt to evolving societal and technological landscapes.

The AutoML platform works alongside the ETD-Hub. It combines an interaction layer based on generative AI, especially large language models, with an AutoML engine, allowing for the creation of AI tools through natural language and speech interaction, with an emphasis on multiple languages. The platform generates AI models that aim to respect the EU legal framework and provides access to various datasets, promoting balanced AI development.

ALFIE’s goal is to make AI development more accessible while encouraging ethical compliance and transparency. By linking policy discussions with technological implementation, ALFIE aims to support responsible AI practices that can be used by a wider range of users and stakeholders.

KInIT contributes to the ALFIE project in two main areas:

AI ethics and regulation: KInIT is responsible for analyzing ethical principles in modern AI solutions. This involves examining existing ethical principles for trustworthy AI, assessing their practical implementation in real-world applications, and suggesting potential improvements to enhance their applicability in various AI contexts. KInIT’s work in this area contributes to the ETD-Hub discussions and policy recommendations.

AI and Data, focusing on Natural Language Processing: KInIT works on creating AI solutions through multilingual interaction in natural language. This includes adapting large language models for translating user queries into inputs to the AutoML engine to enable creation of explainable AI solutions, as well as working on speech-to-text and text-to-speech models to improve platform accessibility. To close the loop, the interaction layer will also explain the solutions created by the AutoML engine in a language users can understand. KInIT’s efforts focus on enabling multilingual interaction with the platform in both major and minor European languages.

Through these contributions, KInIT aims to help ALFIE integrate ethical and societal considerations into practical AI development while working to make AI technology more accessible across different languages. KInIT’s involvement seeks to help connect ethical principles with their application in AI development, contributing to the creation of more responsible and inclusive AI technologies.