The Double Prism
Ideas at the edge of AI, science, and software
-
A summary of design patterns to secure AI agents
A practical summary of design patterns from the paper 'Design Patterns for Securing LLM Agents against Prompt Injections'
-
A Pragmatic Evaluation of Software Engineering AI Tooling
How we evaluated Claude Code, Cursor, and GitHub Copilot across 77 engineers and 165 real tasks to determine productivity impact
-
Rationalizing the AI bubble
An analysis of AI bubble through financial data: examining revenue gaps, circular deals, and whether we're heading for a meltdown or just a price correction
-
Beyond Tokens: The Context-Window Perspective on LLMs, Memory, and Mind
Exploring the bridge between next-word prediction, agent frameworks, and the limits of current LLMs consciousness
-
Launching TheorIA: A Machine-Readable Atlas of Theoretical Physics
If we want AI models to reason about physics, we first need to give them physics they can actually read.
-
Datasets for advancing Theoretical Physics and AI
There is a lack of curated datasets in theoretical physics to train better machine learning models. But what exactly is missing and how can we fill the gaps?
-
Selected ideas from NeurIPS 2024
NeurIPS 2024, the largest AI research conference, provides a glimpse into the next frontiers. Here are some of the most exciting ideas presented.
-
Opening the LLM pipeline
My notes on a great tutorial at NeurIPS 2024 on how to build a Large Language Model, with many practical tips.