Iwona Bialynicka-Birula ⏩
@iwonabb
Followers
1K
Following
3K
Media
215
Statuses
1K
making artificial intelligence super enough for real-world enterprise applications
Redmond, WA
Joined July 2009
Actual footage of an AI using a lot of water per query
0
0
0
Building products on top of foundation models is a completely new discipline that didn't exist a few years ago. What I love about it is that it enables entirely new product categories and involves constant learning as model capabilities evolve every day. A great illustration of
cresta.com
Take a deep dive into how we built Cresta AI Analyst, and all of the lessons learned along the way.
0
0
0
You can tell that a talk is really good if you're simultaneously thinking "this is obvious" and "why did I not realize this before?"
ICYMI, here is my keynote on the semantics of LRM "thinking traces" at #NeurIPS2025 workshop on Multimodal Algorithmic Reasoning. It's a unified view of seven papers we presented at the conference workshops. Special thanks to the attentive audience🙏 👉 https://t.co/PYLd97vBKQ
1
2
6
The future is going to be stranger than anyone has really imagined yet. There are things being realized in frontier alignment research about the fundamental nature of the machines. The danger is not Skynet. The goal is not "super intelligent tools." The coherence that is
109
53
650
We will know that AGI has been achieved when it doesn't take world-class AI researchers 5 minutes to connect a laptop to the projector (still not achieved at #NeurIPS2025)
0
0
9
Today at #NeurIPS2025, we present Titans, a new architecture that combines the speed of RNNs with the performance of Transformers. It uses deep neural memory to learn in real-time, effectively scaling to contexts larger than 2 million tokens. More at: https://t.co/pEuI5oDkkn
57
268
2K
Oh wow, I made this reply and my For You feed instantly switched to nearly 100% NeurIPS/AI content. Finally! I just wish it was more explicitly controllable...
@NeurIPSConf This app is some vibecoded slop. I hope that @NeurIPSConf reveals that it was meant as a secret coding agent benchmark.
0
1
0
Quintessence of why Europe can't have nice things: AI researchers create a LLM optimized for European languages - asked why it's not on HF - can't share it yet because of red tape created by AI act...
@fr3akpl Jeszcze chwilę bo musimy przygotować pełną dokumentację projektu - Ai act.
0
2
3
Why? Does Cursor now want to listen to music while coding?
0
0
2
Happy Birthday, Chat!
today we launched ChatGPT. try talking with it here: https://t.co/uWra8LKFMN
1
0
2
So... they're looking to hire an LLM fine-tuned to mimic an AI platform director? https://t.co/GJ0HgEXsN9
2
0
5
The blue part is expanding as well
0
0
0
AI application layer: There is underinvestment. The potential is still much greater than most realize. 👈💯
Is there an AI bubble? With the massive number of dollars going into AI infrastructure such as OpenAI’s $1.4 trillion plan and Nvidia briefly reaching a $5 trillion market cap, many have asked if speculation and hype have driven the values of AI investments above sustainable
0
0
1
What I'm trying to say, is that we have a whole lot more work to do than this image suggests.
0
0
0
Benchmark results look impressive (and they are!), but the moment a task becomes a benchmark it naturally becomes a measure of narrow intelligence, not of generalization.
1
0
0
To get an idea of where we are with "AGI", give frontier models with maximum test-time compute a screenshot of the NYT Pips puzzle and watch them fail miserably at trying to solve it.
1
0
0
"Did you evaluate this AI agent before deploying it?"
0
0
4
Awesome video about the history of contact centers
💡 CX has transformed from phone support in the 1960s to today’s AI-powered era. At #CrestaWAVE, we shared our vision for what’s next: AI & humans working together to deliver more personal & proactive experiences. ✨ Watch the full keynote on-demand here: https://t.co/V5B3l4UnK9
0
0
4