5 Comments
User's avatar
June Dershewitz's avatar

If a certain dataset is already deemed "important" and it's well-formed enough to be generally useful, by all means use it as input for an AI proof-of-concept. In my experience, that's what I'm seeing.

Expand full comment
Yoni Leitersdorf's avatar

That is directly in line with that I'm seeing that works too!

Expand full comment
June Dershewitz's avatar

I mean, it makes sense.

BTW I feel obliged to mention my own GIGO post here. Everyone who has a data newsletter has to write a GIGO post, it seems. :-) https://measurecraft.substack.com/p/garbage-in-garbage-out-is-a-lame

Expand full comment
Yoni Leitersdorf's avatar

Ha! I remember reading that now. It must have been so ingrained into my brain that I unconsciously used it. I'm like an LLM it seems ;)

Expand full comment
June Dershewitz's avatar

Chicken and egg. I do it, too.

Expand full comment