danandsomehandsome.bsky.social
16 posts
58 followers
232 following
Conversation Starter
comment in response to
post
I’m curious on your thoughts to recent developments in the space in the form of the article: www.scientificamerican.com/article/insi...
And I’m also not a fan of following up a scientific journal article with a magazine article… appreciate your thoughts/thread
comment in response to
post
I’m hoping we start to see more sports and events in the weekday windows (even if not “prime time” evening spots), like TGL tried this winter - honestly I’d like to see them more spread out to not compete with each other
comment in response to
post
I understand that and agree with it. I’m curious if non-live TV plays the same way and decreases the competition for eyeballs during late night windows. Is “prime time” just a sports specific term now? Does it still actually apply to linear tv the same way it used to?
comment in response to
post
How has “prime time” shifted in this era of streaming? How many viewers are accessible at 3 PM ET vs 8 PM ET on Tuesday? DVR, streaming, and the ubiquity of screens have to have changed the standard of what “prime time” really means?
comment in response to
post
Einstein’s puzzles, or zebra puzzles, do exactly this, in my opinion… I think mentioned elsewhere as logic puzzles, but they are a great reading comprehension intro into logic
comment in response to
post
Easy solution: morning baseball
comment in response to
post
It’s on TNT at 7 & 8 PM ET - www.unrivaled.basketball
comment in response to
post
Interesting, I appreciate the input you’ve provided, thank you. Do you have any recommendations on what to read further to better establish my knowledge here?
For example, “limited mostly by the number of features”? Thank you and thank you #mlsky #statsky
comment in response to
post
Yes. It very much is. I can down sample the negative observations, and use other methods to help balance, but at some point the skew should say, “don’t build a model on that data.” I’m looking to understand what that point would be.
It’s a propensity model predicting many populations, some n> 1 M
comment in response to
post
Do you believe that ageism affects younger people in politics as well? And if so, is the age limit (at least 35) on becoming a candidate for president influential?
comment in response to
post
I’d enjoy a breakdown by @carlbergstrom.com here, though I’m betting it’s very similar to their earlier study, but did they do enough to improve the procedure? Based on the paper they attempt to decorrelate, and outside of that weight is a big factor, still no concerning results.
comment in response to
post
I understand. And I apologize if I didn’t make you believe that I understood. I attempted to relate this by saying “a large percentage”, which is subjective, and I’m happy for you to define it. I’m not defending “AI” users, but rather commenting, many application services use them against us already
comment in response to
post
Why does the reviewing party get to be the ones shirking responsibility? I’d hope that reducing requirements of apps would be able to unlock the ability for humans to read these letters of recommendation when provided (cover letters, etc). If a human is reading, the expectation is different.
comment in response to
post
I’d like to argue that a large percentage of letters of recommendation are “formalities”, because models are used to parse and evaluate entire applications. I can’t speak to your “someone you know”, but if a model is reading my app, is it so bad a model write it for me?
comment in response to
post
I appreciate the distinction between AI and ML - but bummed I had to dig for it in a thread… I agree about the bubble to some extent, there are a lot of groups using “AI” (non LLM) to explain where they had been using “ML” (to promote their businesses/ speak to investors) because that’s the buzzword