|
Jeudi, 3 février 2022 DeepMind claims that its artificial intelligence called AlphaCode is as good as an "average human programmer"Posté par Julien Teyssier dans Nouvelles technologiesCommentaires (0) | Rétroliens (0)
AlphaCode has been tested on challenges organized by Codeforces, a competitive coding platform that publishes problems on a weekly basis and establishes rankings for coders, much like the Elo ranking system used in chess. These challenges require in-depth knowledge of algorithms and theoretical concepts of computer science. These are highly specialized puzzles that combine logic, math, and coding expertise.
In an example challenge AlphaCode was subjected to, competitors are asked to find a way to convert a string of random, repeated letters s and t to another string of the same letters by using the backspace key. AlphaCode proceeds to read the problem and proposes the source code of a solution to it. More information in French: DeepMind déclare que son intelligence artificielle dénommée AlphaCode est aussi bonne qu'un « programmeur humain moyen »…
More information in English: Competitive programming with AlphaCode | DeepMind Rétroliens
Rétrolien spécifique pour ce billet
Pas de rétroliens
Commentaires
Afficher les commentaires en
(Vue non groupée | Vue groupée)
Pas de commentaires
Ajouter un commentaire
|
Recherche rapide Catégories Calendrier
Administration du blog |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||