What is your convincing dream scenario for AI – a breakthrough that may fundamentally improve life at a world level?
AI that truly helps people understand complex concepts without feeling silly. Imagine that your doctor's explanation in your blood work doesn’t require a medical conclusion, or legal contracts for humans are written first and the lawyers in second place. Non -simplified, translation.
What is a current project or a breakthrough that you simply are particularly happy with – and what sort of effects, what effects it’s going to have in the actual world?
We principally built a factory after training on the Nebius Ai Studio. The breakthrough isn’t a thing. We discovered how we’ve got massive models (consider 70b+ parameters, MOE architectures) with really long contexts (as much as 128,000 tokens) without having a supercomputer budget. The boring technical part has solved a whole lot of distributed training problems with context parallelism, professionally, a lot. The interesting part: Researchers and firms can now train models on their very own data in a scale that was principally unimaginable six months ago, unless they were Google. We open it as a platform this quarter. Real-World Impact? To be honest, I’m most enthusiastic about what people will construct on that we’ve got not considered. Long context training opens up things like agents who can work with entire code bases, models that understand complete legal documents. “
What is an application for AI that you simply consider that more people should know – something positive that flies under the radar?
Podcast producers who use models which can be finely coordinated within the type of their show to routinely remove filling words, ingredients and uncomfortable breaks. The host still feels like himself, only narrower. Shortens the processing time by 70%.
If you’ve got to decide on a nightmare scenario that keeps you awake at night – whether realistic or speculative – what would it not be and which warning signs should we take a look at today?
People who forget the way to think. The algorithm decided how the AI becomes an excuse, and suddenly no one can explain within the room why a call was made because they outsourced the argument. From now on you may see it when people treat model editions equivalent to gospel.
Who or what do you’re thinking that has the strength to stop your nightmare scenario above?
We all. Attention to the chain of thoughts, the outputs. We do our greatest to avoid AI slop.
What are we not talking about within the AI discussion today – something that you simply consider that it could possibly be very necessary in five years?
That we create a world through which some people use AI and other persons are simply utilized by it
If you look ahead for 10 years, what do you’re thinking that will likely be the largest change in our each day life?
Probably how we take care of it to not know things. At the moment, if you happen to don't know something, have a look and get a solution. In 10 years you’ll receive a probability distribution and have to search out out what to do with it. Maybe that makes us more thoughtful. Maybe it only makes us more anxious. Turn a coin.
What did you encourage you to participate on this AI summit as a spokesman and what message you hope to send the audience?
Every time I see someone who could construct something, he wouldn’t have built or solved an issue before which he recorded is the payment. I’m here, to speak about this page, not concerning the existential debates or the market hype, but the sensible query of how we make powerful tools accessible to individuals who have to unravel interesting problems
Dylan Bristot, product marketing director, Nebius Ai Studio
08 – 09 October 2025
Taets Art & Event Park, Amsterdam
09 – December 10, 2025
Doha exhibition and congress center
