Again, I am asking you for a theory. WHY would strings of text reflecting word distribution in some unknown training corpus be useful for the work that you are doing? >>
@emilymbender
-
AI Training Data Validation and Reliability Concerns
By
–
That said, no, you still haven't answered my question. *Especially* with closed training data sets, to say "Well, it seems to work in the instances we tested" is extremely shaky ground for using a system to produce data for analysis. >>
-
LLM Theory and Data Source Validation
By
–
You didn't answer my question though. What is your theory of what LLMs do and how does that theory motivate your using their output as data for your purposes?
-
Don’t Use AI Art to Depict Unfamiliar Cultures
By
–
Protip — don't use synthetic "art" to illustrate your articles and ESPECIALLY don't do so if you're prompting it to depict a scene from a culture (or like here cultures!) you don't belong to. This is some epic self-beclowning. (h/t @jeffadoctor who found this atrocity.)
-
Concerning prompt engineering results spark worry
By
–
I shudder to think what prompt they put in to get that…
-
GPT-4 limitations for large-scale data automation use cases
By
–
I have no objections to using automation to do large-scale data processing. But it needs to be developed & tested for the use-case. "zeroshot" (ugh) or not, GPT-4 isn't that. You don't even know what its training data is.
-
Corpus Frequencies as Research Data for LLM Analysis
By
–
What is your research question, and why would word sequences reflecting the frequencies in an undocumented corpus be a valuable source of information in answering it? (And no, this isn't about GPT-4 only; that's just the one in question today.)
-
Stochastic Parrot Named AI Word of Year 2023
By
–
Stochastic parrot is @americandialect 's AI related word of the year for 2023! https://
americandialect.org/wp-content/upl
oads/2024/01/2023-Word-of-the-Year-PRESS-RELEASE.pdf
… (AI related WOTY was an ad hoc category this round. Here's hoping there does not continue to be a need for such a category.) w/
@mmitchell_ai @timnitGebru @mcmillan_majora -
Technosolutionism and Structural Inequality: AI Expert Crossover
By
–
Ready for more Mystery AI Hype Theater 3000? Next up: @TheGoodRobot1 crossover episode! @DrEleanorDrage and @KerryAMcInerney join @alexhanna and me to discuss technosolutionism and structural inequality Monday Jan 8, 9am Pacific
-
Basic AI Privacy Protection More Urgent Than Exotic Regulations
By
–
exotic new regulations that should be passed to address AI while they consider exotic problems like p(doom) or whatever, we're not doing some of the maybe basic things we need to do, which is perhaps just address you know basic privacy protection. /End alttext