potatoman22

5 ms·

61.
by potatoman22 2y ago
It's trendy to say "it's horrible to use AI for this" without giving specific reasons. Some reasons it could be good to use AI here: - this can prioritize urgent patients for the severely overworked doctors - medical err
62.
by potatoman22 2y ago
This is how most ML in healthcare is used. It's a fantasy to think we can give all patients equal attention, so it often makes sense to prioritize those most at risk. Of course no model is perfect, but ML can be very beneficial when us
63.
by potatoman22 2y ago
To save a click: "We define cognitive autonomous agents as an entity designed to perform tasks autonomously by combining four key components: goals, reasoning/planning, memory, and tools."
64.
by potatoman22 2y ago
I'll have to use this, thanks for sharing. Isn't it problematic since Gemini isn't representative of a real user, though?
65.
by potatoman22 2y ago
The 82TB Meta trained on is still a lot of textbooks.
66.
by potatoman22 2y ago
It's really deceptive how the title of this post implies causality despite the study not stating a causal link exists. "Coffee reduces risk of T2DM" is much different than the author's conclusion: "Adding sugar or a
67.
by potatoman22 2y ago
For those like me: Cafetière means "French press."
68.
by potatoman22 2y ago
> The OCR returns confidence intervals. If the LLM reasons over OCR that it is not confident about, we flag this in the UI to the end user and ask a human to review before moving forward. This seems helpful, but what if the flagging syst
69.
by potatoman22 2y ago
Totally an anecdote, but my 8gb M1 runs fine with multiple browsers/tabs, VS Code, and Spotify all open. Usually performance is only an issue for me when working with larger ML models. I wonder why others are getting worse performance?
70.
by potatoman22 2y ago
Should doctors also require patient consent to Google disease symptoms?
71.
by potatoman22 2y ago
It reformulates the visit transcription as medical notes. That's the "virtual assistant" part afaik.
72.
by potatoman22 2y ago
There's work in that area, it's sometimes called "accompaniment generation." https://arxiv.org/abs/2301.12662 https://fastsag.github.io/
73.
by potatoman22 2y ago
I've thought about it and don't understand what you're saying
74.
by potatoman22 2y ago
Play chords to transmit more info?
75.
by potatoman22 2y ago
I don't see integrating AI into UI/UX to be incompatible with enabling the user to do as she intends. In fact, I think the thoughtful use of AI could better help users do what they want. It is annoying when "smart" featu
76.
by potatoman22 2y ago
Maybe the lesson should be: a good product can make money even without a moat.
77.
by potatoman22 2y ago
The model might not use the tools every completion, depending on your setup.
78.
by potatoman22 2y ago
It might be the way you're formatting the input. I wonder how they perform when state updates are shared via natural language vs ASCII art vs image
79.
by potatoman22 2y ago
Before, we browsed Tinder or met people IRL. Now, we just text ChatGPT.
80.
by potatoman22 2y ago
I share your frustration. It ought to be different, and some companies are, but embedding ads in chatbots seems inevitable the world that is.
81.
by potatoman22 2y ago
I've had decent luck using some of the reasoning models for this. It helps if you task them with identifying where the points on the graph are first.
82.
by potatoman22 2y ago
Small point but is it doing semantic chunking, or loading the entire pdf into context? I've heard mixed results on semantic chunking.
83.
by potatoman22 2y ago
In my experience: all over. It's the company that teaches them not to screw up the important work.
84.
by potatoman22 2y ago
You probably just get more spam texts since you're replying. Maybe that's a good thing tbh
85.
by potatoman22 2y ago
I see this fallacy a lot in the US. I think it's because of our individualism. Attention and hard-work can't overcome everything, we aren't all-powerful beings.
86.
by potatoman22 2y ago
Why? Past progress =/= equal rate of future progress.
87.
by potatoman22 2y ago
I'd rather have buggy things now than perfect things in a year.
88.
by potatoman22 2y ago
That's great to hear, thanks for the explanation! Super excited to try this out.
89.
by potatoman22 2y ago
How does the token usage compare to vanilla structured output? Many of these libraries do multiple requests to constrain output and measure logprobs.
90.
by potatoman22 2y ago
Woops. I should have done more reading before posting.

More ›