Although it’s the third time Itera is represented at UTmessan, each time feels like a new adventure and an amazing experience. The journey to Reykjavik was challenging; it took us almost two days to reach the city — not only due to a lack of flights from Ukraine but also because of severe weather conditions in Iceland. However, despite these obstacles, it was worth it. Attending UTmessan is not just about witnessing technological advancements; it's about being part of an innovative community and expanding new connections.
The atmosphere of the conference day was warm and welcoming, providing a supportive atmosphere for attendees to engage in discussions and share ideas. With around 1,400 guests present and approximately 8,000 visitors at Harpa the following day — where Reykjavik’s population is about 140,000 — the energy was palpable.
The conference featured 60 lectures across 10 tracks, with AI and cybersecurity taking centre stage. Itera's own Snæbjörn Ingi Ingólfsson officially opened the event with an inspiring keynote that set the tone for what lay ahead. In addition, Odd Hafid Khalifi gave a presentation about our value proposition at Itera — highlighting our strengths and contributions within the tech landscape.
It was exciting to be on stage and share a talk from my own domain expertise; breaking AI chat systems. The presentation was crafted around my deep experience in testing AI applications, particularly Itera’s internal AI platform, Sapience. Testing AI is different from traditional software testing because AI systems learn and generate responses dynamically. To ensure AI is reliable and safe, the following approaches are essential:
- Keeping conversations smooth and maintaining context.
- Detecting and reducing biases to avoid misleading responses.
- Strengthening security to prevent AI from being misused.
- Stress-testing AI with tricky questions, different languages, and large data inputs.
- Adding ethical safeguards to prevent AI from generating harmful content.
We can make AI smarter and safer by continuously improving these testing methods. .jpg)
.png)
-2.jpg)
.png)
%20copy.png)
.png)
.png)
.jpeg)
-1.jpg)
.png)
.png)
.png)
.jpg)
.jpg)
.webp)
.webp)
.webp)
.webp)
.webp)
.png)
.jpg)
