caps lost 3-5 to the blues, the boys looked sloppy, but ovi and kyrou got their first of the season and jenner had another nice one my dog is cute af when he sleeps and cuddles up to me
rob
@bisks.net
🐴🪽 Full fathom five Cornscraper lies; Of his cobb marshmallow-made; Peeps piled to tremendous size: Nothing of it that doth fade, But doth suffer a sea-change Into some great gooey grange
i do not post much about my dog / wife / hockey / work those things loom larger in my life but are idk embodied real-time kinds of things; what is there to say about them
I disagree with the particulars of the study design, and with their bold self-serving proclamations that the study says something general about ai use
i agree with the authors on a number of things - this study has appropriate statistical power - the study would replicate - ai used this way in a learning situation would be a negative for learning
the screenshot is the method of the study my specific criticism is that the study could not find anything meaningful as designed “its dumb” is not my argument but it is the takeaway here, the summary of my argument
yea fair this was too hot
do I think ai is bad for learning, if you use it to answer the question for you? yea, pretty obviously
you can tell by reading it that its just poorly constructed to find anything that generalizes beyond the particular circumstances of the study task itself
but the task in the experiment in the paper is stupid and says nothing about anything interesting
lots of ai in education seems quite bad, the ideas the paper is connected to are most likely correct in many ways re education
I have not seen what that person has said, so have no idea
there is nothing interesting about human behavior or the affect of ai on behavior that could be determined by this study design this is flash in the pan power pose bullshit
but… the study is bad? why would you design the study in this way? what could you possibly find of any interest to anyone? it seems almost perfectly designed to create this exact credulous headline and braindead response
til! gotta say it does bother me that claude is claude so annoying and pedantic abt the explanation tho, gotta add “don’t be a pedant” to the anti-sycophant training
why would anyone post something like this
not that different! but this new hire doesn’t sleep and can make clones of themselves and is cheaper than any human to hire and runs on silicon instead of carbon for better or worse
right its quite weird bc sometimes with just a little nudge they are paragons of taste, I feel like I’m stumbling in the dark wrt steering in this way
I think he clips for yt in ways that models are still somewhat tasteless at? they sometimes have trouble finding the start and stop of a good clip, even with the transcript, they know too much or something? idk, it does not seem like it ought to be out of reach of current models I guess…
the purpose of a model ising is what it doesing
neither heavenbanning nor hellbanning but some weirder alternative universe
