RicardoRei

4 ms·

1.
MindGuard: Open-source safety classifiers for mental health AI(swordhealth.com)
5 points by RicardoRei 9mo ago|1 comments
2.
by RicardoRei 9mo ago
Hi, HN, I'm Ricardo, Head of AI Research at Sword Health. We've been working on AI mental health support for a while now, and one of the biggest challenges we kept running into is how poorly general-purpose safety classifiers work
3.
by RicardoRei 10mo ago
Again, these things don't depend on each other. LLMs have room for improvement (we show that their scores are medium-low on several dimensions). Maybe the average human also has lots of room for improvement. One thing does not necessar
4.
by RicardoRei 10mo ago
Cite the ArXiv paper for now.
5.
by RicardoRei 10mo ago
yep yep. Its something we have to study and its likely we can improve the LLM as a Judge further. Same thing for the patient LLM. We can probably fine-tune an LLM to do a better job at simulating patients. Those two components of our framew
6.
by RicardoRei 10mo ago
Exactly. We don't do claims about humans. But there is room for improvement on current LLMs... For researchers to be able to improve LLMs we first need to know how to evaluate them. We can only improve what we can measure so we studied
7.
by RicardoRei 10mo ago
The results are not meaningless but they are not comparing humans against LLMs. The goal is to have something that can be used to test LLMs on a realistic mental health support. The main points of our methodology are: 1) prove that is possi
8.
by RicardoRei 10mo ago
This is a good point. We have not tested the clinicians but I believe they would not score each other perfectly as we observed some disagreement also between the scores which also reflects different opinions between clinicians
9.
by RicardoRei 10mo ago
The prompts are kept the same for all models. Otherwise comparison would not be super fair. In any case you can check all the prompts in our github repo.
10.
by RicardoRei 10mo ago
Hi HN - I’m the Head of AI Research at Sword Health and one of the authors of this benchmark (posting from my personal account). We built MindEval because existing benchmarks don’t capture real therapy dynamics or common clinical failure mo
11.
New benchmark shows top LLMs struggle in real mental health care(swordhealth.com)
114 points by RicardoRei 10mo ago|157 comments