Went to a tech meetup last night, had a lot of younger developers there.
-
Went to a tech meetup last night, had a lot of younger developers there. Was amazed by how many side projects were essentially the same inventory/POS projects people were doing in xBase back in the '80s.
Only this time the tooling is way less suited to task.
-
Went to a tech meetup last night, had a lot of younger developers there. Was amazed by how many side projects were essentially the same inventory/POS projects people were doing in xBase back in the '80s.
Only this time the tooling is way less suited to task.
@danlyke Ha! That *is* surprising! What on earth are they inventorying!
I would've expected something more like what I've been working on: apps that are, in large part, LLM prompt generators. I could easily see contemporary inventory apps having that as a significant feature for analytics.
All to say, in some ways, I tend to see LLMs as a lazy person's data mining. You know, machine learning from the late 90s/early 00s?
(Mine has more client side tracking/analytics but still...)
-
@danlyke Ha! That *is* surprising! What on earth are they inventorying!
I would've expected something more like what I've been working on: apps that are, in large part, LLM prompt generators. I could easily see contemporary inventory apps having that as a significant feature for analytics.
All to say, in some ways, I tend to see LLMs as a lazy person's data mining. You know, machine learning from the late 90s/early 00s?
(Mine has more client side tracking/analytics but still...)
@danlyke That is, pattern analytics finding coincidences and possible causality as a way for optimizing or detecting... whatever.
-
@danlyke Ha! That *is* surprising! What on earth are they inventorying!
I would've expected something more like what I've been working on: apps that are, in large part, LLM prompt generators. I could easily see contemporary inventory apps having that as a significant feature for analytics.
All to say, in some ways, I tend to see LLMs as a lazy person's data mining. You know, machine learning from the late 90s/early 00s?
(Mine has more client side tracking/analytics but still...)
@elight things like "I got into collectibles from gaming, and I'm working on better software for when I have a booth at cons".
I think everyone's working on LLM prompt generators for their day job these days.
-
@danlyke That is, pattern analytics finding coincidences and possible causality as a way for optimizing or detecting... whatever.
@elight I think that generally requires a large enough data set to do something with.
This is the problem I ran into building face recognition for my photo library: I just don't have that many pictures, and especially not enough carefully aligned portraits, to build real training sets.
So anything I do alone is constrained by outside sources.
-
@elight I think that generally requires a large enough data set to do something with.
This is the problem I ran into building face recognition for my photo library: I just don't have that many pictures, and especially not enough carefully aligned portraits, to build real training sets.
So anything I do alone is constrained by outside sources.
@danlyke Ah! For me, it's data supplied by the user, looking for behavioral patterns over time.
-
@elight I think that generally requires a large enough data set to do something with.
This is the problem I ran into building face recognition for my photo library: I just don't have that many pictures, and especially not enough carefully aligned portraits, to build real training sets.
So anything I do alone is constrained by outside sources.
@danlyke And here I just want someone to make a modern TODO specifically designed to take on the cognitive load of managing too many TODOs.
I love TODOs and I also hate them so very very much.
-
undefined oblomov@sociale.network shared this topic