Jeremy Berman's agent scored 96% on ARC-AGI-3 by compiling its reasoning into executable code. It's a powerful approach—but real-world problems like business are a lot harder to hill-climb than ARC puzzles.
What if you had an AI agent that took in your voice notes and helped you draft an algorithm-optimized banger, in your voice, to communicate what you want to as wide an audience as possible?
A demo for Albany Can Code: connect a Cursor Cloud Agent to a Hetzner VPS and ship a Jackbox-style game to the internet in under ten minutes.
To get quality outputs from an AI agent, have it make a detailed study of examples of quality outputs first
A short story
... and this is the least conscious they'll ever be
A technical overview of the core data structures in transformer-based language models: the embedding matrix, stacked transformer blocks, and language modeling head.
My mental model was that token embeddings represent points in n-dimensional space, and distance between points represents semantic similarity. Turns out that's not quite right.
A recording of my DevFest DC presentation on using Simon Willison's LLM tool to interact with large language models from the command line. Learn about templates, embeddings, semantic search, structured data extraction, and how to chain LLM with everyday CLI tools.
New IMF API, new IMF API R library — part of the econdataverse initiative
At the July 5, 2025 Coders' Colaboratory mini-hackathon in Latham, NY. Developers had just one hour to find a cool use case for the `llm` command-line tool, created by Simon Willison.
A demo presentation I gave on Simon Willison's incredible `llm` command-line tool at the Coders' Colaboratory mini-hackathon in Latham, NY on July 5, 2025.
All the kids are talking about Claude Code. Maybe you've heard of it. But did you know you can use it to write R code right inside of RStudio? No more messing with Cursor to try to get it to play n...
Announcing a beautiful new documentation website for `imfp`, the leading Python tool for working with International Monetary Fund data
Google's "NotebookLM" generates amazing "deep dives" on any topic you can listen to during your commute
Combine R's powerful features for working with data with Cursor's powerful features for editing code
Learn how to avoid the mutable default argument problem in Python and how to properly initialize mutable objects as default arguments. Understand the behavior of mutable and immutable objects and how they affect your code.
Discover a unique Python quirk: logical operators returning the last evaluated operand. Learn how this behavior affects code execution, enables clever 'code golf' tricks, and provides intuitive ways to handle default values. Explore examples with 'and' and 'or' operators that might surprise even experienced Python developers.
Explore the fascinating world of XOR encryption! Learn how this simple logical operator creates perfectly balanced output, making it a powerful tool for secure data encryption. Discover why XOR outperforms AND and OR in cryptography, and experiment with an interactive demonstration to see these operators in action.
A prompting toolkit for fiction writers
Overestimate the economy and gaslight consumers at your own political peril
Quickly install and run an open-source Llama model with vision capabilities on your personal Mac or Windows PC
Use ChatGPT plugins to quickly find available domain names for your website
Elicit, by Ought Inc., feels like browsing PubMed, but with more power and a little less headache
Could AI infer a "grammar" of consciousness from our language the same way it has inferred a "grammar" of reasoning?
Language models are statistical representations of human thought-space, with profound implications for philosophy of mind
If you've studied an asset and think it's too cheap, don't fool around with lowball bids or you'll expose yourself to adverse selection
252 global economic databases with tens of thousands of indicators, accessible at your convenience
Let's create a Jupyter notebook for generating stunning cover images and writing engaging tweets!
By massively lowering software development costs, OpenAI's large language model will kick off a small-business renaissance
One sector made the lowest lows from pre-pandemic levels... and also the highest highs
Accounting for time effects in forecasting 10-year US Treasury yield and S&P 500 P/E ratio from PCE inflation rate
5 Simple Steps to Using AI and Python for Reproducible Data Visualization
Playing pun games with ChatGPT
What each of the major AI image gen tools is good for and how to get started using it
A newsletter on staying employed in knowledge work through massive technological disruption in the age of AI
Copyright © Christopher Carroll Smith 2024