Kempelen Institute of Intelligent Technologies

We've achieved many things that seemed almost impossible 2.5 years ago when we founded our institute. However, our family and friends often have no idea what KInIT actually is 🤔 So we decided to explain who we are and what we do in a video. 👋 Get to know some of our brilliant colleagues and find out about their daily activities:

🫵Calling all AI researchers! Participate in a study aimed at identifying and addressing challenges within the AI field 🤖 Being the AI-on-Demand Platform a community-driven channel, it is essential to understand the needs and challenges of researchers and the AI research process itself 🧑💻 This can help us to develop targeted solutions and support services and tools to increase the impact and success of AI research and innovation in Europe. Here's how your input can make a difference: ✅ Assess the severity of challenges faced by researchers. ✅ Initiate a dialogue to address these challenges. ✅ Collaborate on solutions through the AIoD platform. Your participation will be anonymous and will take no more than 10 minutes 📝Click here to access the questionnaire: https://lnkd.in/dAuE8GTi

Don't miss out on our CEO Maria Bielikova as a speaker during SAPIE Forum 2024 ✨This event celebrates the creative power of innovation and demonstrates the strength and potential of the local innovation ecosystem. The forum will focus on various pressing topics of today, including Europe’s ambition to become a global innovation powerhouse, the potential of AI to accelerate business growth of European SMEs or current trends in cybersecurity 💻 Find out more: https://www.sapieforum.sk/

KInIT can't be missing from European Researcher's Night on 27.9.2024! 🌍✨ Mark your calendars and get ready for a day full of innovation and science. More exciting details coming soon—stay tuned! 🔬💡

"Advising AI assistant: ethical risks of Oura smart ring" is the name of a new research article recently written by 👉 Michal Gladiš in the cooperation with our researchers Matúš Mesarčík and Natália Slosiarová. 👨💻👩💻 In this article, they delved into the implications of European AI regulation on wearable devices, which offer personalized monitoring and concluded an assessment of their impact on fundamental rights. Those devices are classified as high-risk and therefore subjects to stringent conformity assessments. It is crucial to recognize the ethical concerns related to AI assistants. We believe the research findings spark meaningful conversations about the impact of AI on society. To read the article click here: https://lnkd.in/d6u8RD7Q

Last week we held at KInIT our very first DisAI Project Summer School! 🙌 The event was full of insightful lectures on topic of trustworthy, multilingual, and multimodal articifial inteligence. It was an incredible gathering of experts and researchers exploring AI models and technologies. Highlights of the week included also a memorable boat trip to Devin Castle and a delightful dinner in the city center. Thank you to everyone who participated and made this event a success!

🤖 AI-Auditology, our new research project, aims to uncover how social media algorithms contribute to the spread of disinformation 🔍 In our previous research, which won the best paper award at the RecSys 2021 conference 🏆 we analysed how disinformation content is prevalent in YouTube's recommender system and search. We are now pleased to announce that we can build on this success 🎥 The AI Auditology project will fundamentally change the way we audit the inner workings of social media algorithms. Its main groundbreaking idea is the application of model-based design to support or even automate individual steps in the auditing process 🛠️ The project serves a noble purpose and will have a major positive societal impact on the online media environment. Read more on our website: https://lnkd.in/da4-wiZv

Last Monday we held our first Natural Language Processing Day! ✊ This event connected experts with students and offered new ways to explore the exciting world of AI. 🙌 The highlights of the day included insightful sessions on neural networks and foundational approaches and hands-on workshop for language model fine-tuning. Thank you to all the participants and speakers who made this event a success! We look forward to future collaborations. Thanks our partners: Nadačný fond Telekom (nadacnyfondtelekom.sk) pri Nadácia Pontis / Pontis Foundation a The Spot

Exactly four years ago, the journey of the Kempelen Institute of Intelligent Technologies began, and today, we proudly celebrate our 4th anniversary! 🚀 Over these years, we've combined the expertise of our academic researchers and educators with the innovative companies, bridging the gap between research and business. Our growth has been incredible, and today, our KInIT family has grown to 92 dedicated members, all united in our mission to advance intelligent technologies! 🥳 We want to hear from you! 🎉 Feel free to share your wishes for KInIT in the comments.

Large language models are currently used for many NLP tasks. However, they are especially sensitive to the effects of randomness, such as the format of the instruction or the in-context examples used. When comparing and benchmarking these models, a small perturbation may lead to a completely different ranking. In our recent paper, which was just accepted to the prestigious ACM Computing Surveys journal, we provide a comprehensive survey of papers that address the sensitivity to the effects of randomness, from its investigation to design of mitigation strategies that can deal with these effects. We are very happy that the paper, which was possible through the TAILOR Network of Excellence Centres on Trustworthy AI, DisAI Project and vera.ai EU-funded projects, was a success and hope it will be beneficial for researchers interested in using language models in low-resource settings! Read more on our website: https://lnkd.in/d9ds3fMx