DeepMind Ai belajar fisika dengan menonton video yang tidak masuk akal
DeepMind has developed an AI software writing tool that rivals the average human coder. The results show that a more world-centric view can give AI a set of general and adaptable capabilities. It eliminates the need to separately learn about an apple on a tree, versus an apple in your kitchen, versus an apple in the trash, providing learning efficiency.
Mark Nixon from the University of Southampton, UK, says the work could lead to new AI research directions and even offer clues about human vision and development. “It means they are using an architecture that others may not be able to utilize,” he said.
Read More: DeepMind telah membuat AI penulisan perangkat lunak yang menyaingi Coder Human Piloto rata-rata mengatakan hasilnya menunjukkan bahwa pandangan dunia-sentris dari dunia dapat memberikan AI serangkaian kemampuan yang lebih umum dan mudah beradaptasi.
Anda tidak perlu belajar tentang apel di pohon, versus apel di dapur Anda, versus apel di sampah.
Ini memberikan efisiensi belajar.Mark Nixon di University of Southampton, Inggris, mengatakan pekerjaan itu dapat mengarah pada jalan baru penelitian AI, dan bahkan dapat mengungkapkan petunjuk tentang visi dan pengembangan manusia.
Itu berarti mereka menggunakan arsitektur yang mungkin tidak bisa digunakan orang lain, katanya.
Chen Feng di Universitas New York mengatakan temuan ini dapat membantu menurunkan persyaratan komputasi untuk pelatihan dan menjalankan model AI.
Manfaat menggunakan representasi objek-sentris, alih-alih input visual mentah, membuat AI belajar konsep fisik intuitif dengan efisiensi data yang lebih baik.
DeepMind has developed an AI software writing tool that rivals the average human coder. The results show that a more world-centric view can give AI a set of general and adaptable capabilities. It eliminates the need to separately learn about an apple on a tree, versus an apple in your kitchen, versus an apple in the trash, providing learning efficiency.
Mark Nixon from the University of Southampton, UK, says the work could lead to new AI research directions and even offer clues about human vision and development. “It means they are using an architecture that others may not be able to utilize,” he said.
