vrli.bsky.social

vrli.bsky.social

@vrli.bsky.social

6 followers4 following5 posts

vrli.bsky.social
@vrli.bsky.social

Charts and graphs help people analyze data, but can they also help AI? In a new paper, we provide initial evidence that it does! GPT 4.1 and Claude 3.5 describe three synthetic datasets more precisely and accurately when raw data is accompanied by a scatter plot. Read more in🧵!

View full thread
vrli.bsky.social
@vrli.bsky.social

🧵(4/5) What role could incorporating charts play in AI-driven data analysis workflows? Visualization is essential for humans to synthesize complex information––can AI shed light on human psychology or the underlying principles of visual knowledge discovery? And many more!

vrli.bsky.social
@vrli.bsky.social

Charts and graphs help people analyze data, but can they also help AI? In a new paper, we provide initial evidence that it does! GPT 4.1 and Claude 3.5 describe three synthetic datasets more precisely and accurately when raw data is accompanied by a scatter plot. Read more in🧵!

vrli.bsky.social
@vrli.bsky.social

🧵(2/5) Comparison with two baselines––providing a blank chart and a chart with mismatched data––shows that the improved performance is due to the content of the charts. AI systems benefiting from visualization raises many intriguing questions:

vrli.bsky.social
@vrli.bsky.social

🧵(3/5) How do graphical variations in graphs affect our results? Could there be a new body of scholarship in visualization for AI-oriented chart design? We use default Matplotlib scatterplots––how might visualizations be optimized for AIs?

vrli.bsky.social reposted

Naomi Saphra
@nsaphra.bsky.social

🚨 New preprint! 🚨 Everyone loves causal interp. It’s coherently defined! It makes testable predictions about mechanistic interventions! But what if we had a different objective: predicting model behavior not under mechanistic interventions, but on unseen input data?

You are up to date.