Paul Sufka MD
@psufka.bsky.social
ChatGPT 5.1 (updated by OpenAI yesterday)
November 14, 2025 at 2:01 AM
ChatGPT 5.1 (updated by OpenAI yesterday)
I did a talk on AI recently. Here are the slides and blog outline if you want to see — paulsufka.com/ai/
The Current State of AI in Rheumatology — Virginia Society of Rheumatology 2025 Annual Meeting.
Objectives Define the major types of artificial intelligence relevant to healthcare Describe current and emerging applications of AI in rheumatology Evaluate the limitations, ethical concerns, and …
paulsufka.com
November 5, 2025 at 11:04 PM
I did a talk on AI recently. Here are the slides and blog outline if you want to see — paulsufka.com/ai/
I know OpenAI is doing some medicine specific work but not sure about Anthropic (maker of Claude)
October 24, 2025 at 11:45 AM
I know OpenAI is doing some medicine specific work but not sure about Anthropic (maker of Claude)
Copilot can barely do Excel correctly and they’re both from Microsoft.
Try ChatGPT-5 and/or Claude Sonnet 4.5.
Might take some prompt engineering either way.
Try ChatGPT-5 and/or Claude Sonnet 4.5.
Might take some prompt engineering either way.
October 24, 2025 at 11:45 AM
Copilot can barely do Excel correctly and they’re both from Microsoft.
Try ChatGPT-5 and/or Claude Sonnet 4.5.
Might take some prompt engineering either way.
Try ChatGPT-5 and/or Claude Sonnet 4.5.
Might take some prompt engineering either way.
This type of error occurs if you use a model with too small of context window.
If it was a small/proprietary model, I wouldn’t expect it to handle this task.
Any idea what model was used?
If it was a small/proprietary model, I wouldn’t expect it to handle this task.
Any idea what model was used?
October 24, 2025 at 1:01 AM
This type of error occurs if you use a model with too small of context window.
If it was a small/proprietary model, I wouldn’t expect it to handle this task.
Any idea what model was used?
If it was a small/proprietary model, I wouldn’t expect it to handle this task.
Any idea what model was used?
Vera health offers CME already and is very good.
October 8, 2025 at 11:16 PM
Vera health offers CME already and is very good.
OpenEvidence or Vera Health >> UTD.
I find Grok and GPT5 really good at medical questions as well.
I find Grok and GPT5 really good at medical questions as well.
October 1, 2025 at 9:16 PM
OpenEvidence or Vera Health >> UTD.
I find Grok and GPT5 really good at medical questions as well.
I find Grok and GPT5 really good at medical questions as well.