I was thinking this without feeling confident i understood who was estimating the intelligence on the y-axis.. you, members of each field, or perhaps an LLM caught in a strange loop evaluating its intelligence versus members of each field
If not, this could make LLMs seem more intelligent within the CS field.
Didnโt take your post as an attack. But I did make an observation without connecting it to a concrete mechanisms which could seem defensive. Let me go further. I was mulling internally if code writing is susceptible to cognitive bias in the same way as natural language.
LLMs pick up cognitive biases from training piles but arenโt able to correct for them without human intervention.
๐ช๐ต๐ฎ๐'๐ ๐๐ต๐ฒ ๐ฝ๐ฟ๐ถ๐ฐ๐ฒ ๐ผ๐ณ ๐ฎ ๐ฑ๐ฒ๐ฐ๐ถ๐๐ถ๐ผ๐ป ๐๐ผ ๐๐ผ๐?
There is no market for the decisions that you face on a daily basis.
๐๐ต๐ฟ๐ผ๐ป๐๐น๐๐ ๐๐ ๐๐ด๐ฒ๐ป๐๐ consistently beat some of the most liquid and well researched markets in the world.
So why not give them a try?
open.substack.com/pub/jeremyol...
๐ ๐ฐ Want to try it on your bracket?
Use the code MCPMADNESS to get 10% off your first month of any of our subscriptions. Offer good for the first 100 users (or until April 7).
#MarchMadness #MarchMadness2025 #MCPMadness #MCP @anthropic.com
๐ ๐๐ต๐ฟ๐ผ๐ป๐๐น๐๐ ๐๐ฌ.๐ฌ.๐ญ๐ญ ๐ถ๐ ๐ผ๐๐
๐ค New agent! ๐๐ถ๐ป๐ฎ๐ฟ๐๐ฃ๐ฟ๐ฒ๐ฑ๐ถ๐ฐ๐๐ผ๐ฟ provides robust probability estimates from multimodal inputs for questions with binary outcomes.
๐ Supports Image, Text, and PDF uploads, up to 35MB per request
๐ฌ Demo using BinaryPredictor for NCAA Bracket picks
youtu.be/7FN_zuZ6Dtg?...
๐ March Madness is right around the corner. So we did what any self-loathing IU fan would do and used Chronulus AI to predict the demise of our team.
claude.ai/share/b64a5a...
#MarchMadness #IUBB @indianambb.bsky.social @mcuban.bsky.social
What if city zoning requirements extended to topological equivalence
๐ฆย Chronulus on Claude: Forecasting NYC Ferry Ridership
Youtube: youtu.be/iOPVutyewW0?...
Our MCP server is now open source and available for general use
github.com/ChronulusAI/...
๐ New forecasting chat with Chronulus forecasting agents using @anthropic.com's new Claude 3.7 Sonnet to synthesize inputs to our Chronulus agents and then plot the predictions together with historical data.
๐ Link to full Claude 3.7 Sonnet chat:
claude.ai/share/3bb21a...
When will the LA wildfires be contained? We can predict it, but should we?
chronulus.substack.com/p/when-will-...
Our Forecasting and Prediction Agents are game changers for the #forecasting and #timeseries communities
Here's an example of predicting foot traffic between the NYC 7 Line and a new Light Rail line that does not yet exist.
๐
๐ง๐ฟ๐ฎ๐ป๐๐ฝ๐ผ๐ฟ๐๐ฎ๐๐ถ๐ผ๐ป - ๐ฃ๐ฟ๐ผ๐ท๐ฒ๐ฐ๐ ๐๐ฒ๐ฎ๐๐ถ๐ฏ๐ถ๐น๐ถ๐๐
docs.chronulus.com/0.0.8/exampl...
We've been able to do this for quite a while, but didn't realize anyone was that interested in getting high latency results on sensor data.
We love the verification process on @bsky.app. It's a breeze when you own a domain.
Followed the steps here and we're off ๐
bsky.social/about/blog/4...
@mtainfo.bsky.social
We have a python SDK on the way for conceptual forecasting.
A concept could be anything that has yet to realize data. Things like new projects, new services, or a project that you'd like to estimate the impact of or iterate over scenarios.
Example: Connecting NYC subways to a new rail line.
