Integrated AI: Dataset quality vs quantity via bonum (GPT-4 and beyond)

Download original article Integrated AI: Dataset quality vs quantity via bonum (GPT-4 and beyond) (PDF)
Note: Text provided here for indexing only, please download the PDF above for formatting as intended.

Alan D. Thompson
June 2021

I acknowledge the complex work undertaken at OpenAI, Quickchat.ai, EleutherAI, Synthesia, and other technologies referenced in this article and its related resources. Importantly, given the speed of AI development, this article should be considered superseded within 24 months of its initial release in June 2021. Revisions are to be expected. Correspondence concerning this article should be addressed to Dr Alan D. Thompson, Life Architect, Australia, 2021. https://lifearchitect.ai/contact

PROFESSOR LOUISE BANKS
Suits. Honour, flowers… Colonel, those are all tile sets in Mahjong. God, are they using a game to converse with their heptapods?

COLONEL
Maybe. Why?

PROFESSOR LOUISE BANKS
Well, let’s say that I taught them chess instead of English. Every conversation would be a game, every idea expressed through opposition, victory, defeat. You see the problem? If all I ever gave you was a hammer…

COLONEL
Everything’s a nail.

— (Arrival, 2016)

 

Innovations like Artificial Intelligence (AI) and neural lace are already here, even if they are not yet in most of society’s field of vision (Thompson, 2020). The world of integrated AI supplementing and replacing our intelligence is moments away (Thompson 2021a; 2021b).

Update, October 2021: This paper has been removed pending a new revision…