DeepMind builds AI 'Gato' that is quite close to human level, gameplay, chat, robot arm operation, etc. are possible
that defeated the Go world champion and forced him to retire, has built ' Gato ', which is a single generalization agent that goes beyond the realm of text output. Based on the context, Gato can decide what to do, such as outputting text, moving joints, or pressing a button.
DeepMind, known for the Go AI 'AlphaGo'
A Generalist Agent
DeepMind's new AI can perform over 600 tasks, from playing games to controlling robots | TechCrunch
DeepMind researcher claims new AI could lead to AGI, says'game is over'
Scott Reed, who works on Gato's project at DeepMind, told news site TechCrunch: 'Most AI systems today run on one task at a time or in a small domain, but what's important about Gato is Primarily, it's where a single model of agent can perform hundreds of completely different tasks, such as controlling robots, giving explanations to images, and chatting. '
According to DeepMind, Gato has been trained to perform 604 tasks, including analyzing images for descriptive text, interacting, controlling robotic arms to stack blocks, and playing Atari games. It is said that there is.
The news site, The Next Web, describes Gato as 'the most impressive all-in-one machine learning kit ever,' and 'Gato is different from general-purpose artificial intelligence (AGI) , which thinks things like humans. It makes me worried that I will never achieve general-purpose artificial intelligence. '
DeepMind's new Gato AI makes me fear humans will never achieve AGI
In response, DeepMind Research Director Nando de Freitas said, 'Now it's all about scale! The game is over! Make these models bigger, safer, more computationally efficient, and faster to sample. With smarter memory, more modalities, innovative data, and on / offline, solving these scaling challenges will enable general-purpose artificial intelligence. ' ..
Someone's opinion article. My opinion: It's all about scale now! The Game is Over! It's about making these models bigger, safer, compute efficient, faster at sampling, smarter memory, more modalities, INNOVATIVE DATA, on / offline,… 1 / N https://t.co/UJxSLZGc71— Nando de Freitas ????️ ???? (@NandoDF) May 14, 2022
Solving these scaling challenges is what will deliver AGI. Research focused on these problems, eg S4 for greater memory, is needed. Philosophy about symbols isn't. Symbols are tools in the world and big nets have no issue creating them and manipulating them 2 / n— Nando de Freitas ????️ ???? (@NandoDF) May 14, 2022
in Note, Posted by logc_nt