Об основных идеях статьи Лекуна:
Melissa Heikkilä, Will Douglas Heaven. Yann LeCun has a bold new vision for the future of AI. MIT Technology Review, June 24, 2022. https://www.technologyreview.com/2022/06/24/1054817/yann-lecun-bold-new-vision-future-ai-deep-learning-meta/
(популярный пересказ, комментарии Йошио Бенгио, других ведущих экспертов и самого Яна Лекуна)
"The idea of cognitive architectures, especially ones inspired by the brain, has been around for decades. So have many of LeCun’s ideas about prediction using models with different levels of detail. But when deep learning became the dominant approach in AI, many of these older ideas went out of fashion. “People in AI research have kind of forgotten about this a little bit,” he says.
What he has done is taken these older ideas and rehabilitated them, suggesting ways that they can be combined with deep learning. For LeCun, revisiting these out-of-fashion ideas is essential, because he believes the two dominant approaches in modern AI are dead ends."
"If nothing else, he wants to convince people that large language models and reinforcement learning are not the only ways forward.
“I hate to see people wasting their time,” he says."
Melissa Heikkilä, Will Douglas Heaven. Yann LeCun has a bold new vision for the future of AI. MIT Technology Review, June 24, 2022. https://www.technologyreview.com/2022/06/24/1054817/yann-lecun-bold-new-vision-future-ai-deep-learning-meta/
(популярный пересказ, комментарии Йошио Бенгио, других ведущих экспертов и самого Яна Лекуна)
"The idea of cognitive architectures, especially ones inspired by the brain, has been around for decades. So have many of LeCun’s ideas about prediction using models with different levels of detail. But when deep learning became the dominant approach in AI, many of these older ideas went out of fashion. “People in AI research have kind of forgotten about this a little bit,” he says.
What he has done is taken these older ideas and rehabilitated them, suggesting ways that they can be combined with deep learning. For LeCun, revisiting these out-of-fashion ideas is essential, because he believes the two dominant approaches in modern AI are dead ends."
"If nothing else, he wants to convince people that large language models and reinforcement learning are not the only ways forward.
“I hate to see people wasting their time,” he says."