Export your statements. Open your bank app. Download the last 12 months of every checking and credit card account as a CSV.
Most people have never done this. It takes 4 minutes. Upload all the files to Claude or ChatGPT in one chat.
@aihighlight
-
Using LLMs for personal financial data analysis
By
–
-
Analysis of o1 model performance and prose quality
By
–
The regression from o1 days is something definitely worth the note. Capability went up, prose discipline went down.
-
Comparing OpenAI’s Research and Consumer Product Engineering Gap
By
–
This is probably true for research talent specifically. The product and engineering gap with OpenAI on consumer side is a different story.
-
The Tradeoff Between Traditional Coding and AI-Generated Code
By
–
Stack Overflow gave you an answer you could read and understand. AI gives you code you can run without understanding it. Different tradeoff.
-
Limitations of LLM-based content filtering
By
–
A filter that works only when the model reads the title is not really a filter.
-
OpenAI’s Strategy for Mainstream AI Adoption
By
–
Natural conversation and useful memory are the two things that make people recommend a product to someone who doesn't care about AI. That's the market OpenAI is going after here.
-
Comparing AI coding model performance in practical tasks
By
–
Two days of failure on a $200 plan followed by a one-shot Codex solution on a $100 plan is a brutal side by side.
-
Discussion on AI Model Autonomy and Reliability Issues
By
–
The autonomy issues with 4.7 probably have something to do with this. Hard to trust a model that goes off script.
-
Impact of hallucination reduction on AI use case viability
By
–
Hallucinaton reduction at that scale would change which use cases are actually viable, not just which ones work most of the time.
-
Claude’s unique problem-solving capabilities
By
–
The fact you're always learning something new from it suggests Claude is consistently approaching problems differently than you would have.