If you like to create datasets, you shouldn’t be in a competition to do it. You should be paid to do it, and paid for access to; if you’re a large organization.
AI Science also shouldn’t be a workshop or a thing companies and conferences let people do maybe.
Also I’d suggest a dataset of hallucinated caselaw which will likely be a study in AI unraveling the foundational trust of an entire industry that glues down nearly all the other industries and geopolitics
Science getting better over time but less so for society was always a given, even the field of General Semantics created by the work of Alfred Korzybski was based on that apparent universal law.
That time seems to be over, because of AI, because of Academia’s race to profits, and through a rush to bring these tools to science. Which I am well acquainted with, likely more than can be imagined.
I am appalled, that We have invited large lying models into truth requiring fields. We have created a new classification of noise that neither Claude, nor Shannon, Norbert or I can analyze.
And this will turn AI that could have been used for Science, into a Mad Science. The congregating in large groups unmasked during a pandemic just adds validity to my theory that Science has peaked.
What we’re doing now is mostly not Science.
Prestigious science conferences are no exception.