Head of Telefónica Scientific Research. Reinforcing the Learning in algorithms, work, life. I'm all about the delta. BJJ and snowboard enthusiast.
Barcelona
Joined July 2014
- Tweets621
- Following61
- Followers297
- Likes206
Ioannis Arapakis retweeted
📢 CoEvolution at IETF 126!
Our partner ISI represented the CoEvolution consortium at the IETF 126 – SCITT Working Group in Vienna, presenting the project and its AIMBOM specification.
🔗 Read more: coevolution-project.eu/coevo…
#CoEvolution #IETF126 #SCITT #AIMBOM #HorizonEurope
Asked #ChatGPT and #Claude the same question about glucosamine safety. Only one flagged a June 2026 Nature Metabolism paper linking it to 25% faster Alzheimer’s progression. @AnthropicAI found it. @OpenAI didn’t. In health queries, that’s not a small miss. #AI
Ioannis Arapakis retweeted
ISI joined the “AI Enables Cybersecurity” meetup hosted by COSMOTE TELEKOM at the OTE Headquarters in Athens.
With 100+ cybersecurity professionals & industry experts attending, we proudly showcased #CoEvolution and engaged in inspiring AI-driven security discussions.
#EUProjects
Interview with DEEPSYNC on AI that can sense & adapt to people, e.g., SYMBIOTIK's neural adaptive interfaces to detect stress/cognitive overload for better real-world decisions, and why EIC Pathfinder is vital for low-TRL, high-gain breakthroughs.
youtube.com/watch?v=XfAam5xz…
#TID
Ioannis Arapakis retweeted
If you express any of this to a modern "therapist" they will immediately put you on a cocktail of psychoactive drugs to chemically lobotomize you
Ioannis Arapakis retweeted
Excited to share that our paper "Efficient and Effective Adaptation of Multimodal Foundation Models in Sequential Recommendation" is accepted at IEEE TKDE! 🎉📷 Joint work with @Solis_xurige @alexk_z @iarapakis @joemonj.
Yesterday we had the pleasure of welcoming @Diego_Perino (Meta) to our new Pocket Auditorium 🎙️He shared fascinating work on Responsible GenAI & AI Judges, from Llama2 safety to LlamaGuard V1. A great reminder of the importance of trust & responsibility in AI @TEFresearch
Ioannis Arapakis retweeted
CoEvolution is at #CyberHOT in beautiful Chania, Crete!
Our team is proudly showcasing the project with a roll-up banner, flyers, and a dedicated presentation.
Great networking & deep discussions on the future of cybersecurity in Europe! 🔐🌍
#Cybersecurity #HorizonEurope
Ioannis Arapakis retweeted
AI isn't secure until it's secure from start to finish.
CoEvolution = full-lifecycle security, trust & robustness for AI supply chains.
🌐 coevolution-project.eu
#SecureAI #EndToEndAI #CoEvolution #EUFunded
Ioannis Arapakis retweeted
What if AI had a blueprint?
With AI MBOM, CoEvolution gives models a structure for transparency & compliance.
🌐 coevolution-project.eu
#MBOM #AItransparency #CoEvolution #EUFunded
Hey @amazon, @Apple, @Google, @Meta, time to rethink user attention! Clicks/views are outdated for today’s SERPs. At #SIGIR25 we present AdSight: a scalable, SoA method using mouse trajectories + Transformers. Preprint: arxiv.org/abs/2505.01451 #AI #TelefónicaScientificResearch
Ioannis Arapakis retweeted
Want the latest on AI security, trust & robustness?
📬 Subscribe to CoEvolution’s #newsletter for exclusive updates & insights.
👉 coevolution-project.eu/newsl…
#AInews #SecureAI #CoEvolution #EUFunded
Ioannis Arapakis retweeted
Instead of seeking feedback on how you did, it’s more effective to seek advice on how you can do better.
Proud of Telefónica Scientific Research at #ICLR25! Two posters accepted (26.65% rate), strong interest in our work on gaze-based AI alignment. Our neuroscience-inspired vision is more relevant than ever. #TelefonicaScientificResearch #TID #AI #Symbiotik #EIC #Pathfinder
Ioannis Arapakis retweeted
AI needs transparency.
CoEvolution introduces AI MBOM, a universal way to describe and audit AI models.
🌐 coevolution-project.eu
#AIethics #AIgovernance #CoEvolution
Ioannis Arapakis retweeted
Our greatest regrets are rarely failures to reach our goals. They're usually failures to uphold our values.
Few people wish they'd been more ambitious. Many wish they'd been more courageous and generous.
Success lies in building a career. Fulfillment lies in building character.
Big congrats to Ángela López Cardona & Sebastian Idesis on our latest IJCNN publication! We review how cognitive signals, esp. eye tracking, enhance language & image understanding in LLMs & VQA. Preprint: lnkd.in/erHm9NUx
#AI #NLP #eyetracking #LLMs #IJCNN #TID
Honored to speak at #EICSummit2025 on AI innovation in Europe. Key message: If Europe wants to shape AI with its values, it must lead, not just regulate. Innovation is the entry ticket. The future? Open source, embodied AI, & a culture shift in how we build.
#AI #TID #eic2025
