Fiche du document
5 octobre 2023
- handle: 10670/1.9346j9
info:eu-repo/semantics/openAccess
Sujets proches
CasketsCiter ce document
Arthur Sarazin, « LLMs are not necessarily black boxes », Dataflow, ID : 10670/1.9346j9
Métriques
Partage / Export
Résumé
Somehow we might think of LLMs as black boxes giving smart answers, out of the blue. Its users are considered by some as lazy individuals not making any effort to perform their own actions. Is that the only use we can make of LLMs ? As artefacts, are LLMs essentially black boxes creating lazy users ? Obviously, the answer is no and I can prove it with a small app I just designed (thanks to this wonderful tuto by Avra) Chaining different lines of reasoning It would be inappropriate to say tha...