Research scientist Nando de Freitas from DeepMind recently said "the game is over" when solving the most difficult challenges in the race to execute artificial general intelligence or AGI.
A Mail Online report specified a British firm owned by Google, DeepMind, may be "on the verge" of attaining artificial intelligence.
Essentially, AGI is a program or machine that can learn or understand any intellectual task that a human being can do even without training.
De Freitas, a machine learning professor at Oxford University, said the quest for scientists has now scaled up AI programs, such as with more data and computer power to develop an AGI.
Gato AI Agent
Earlier this week, DeepMind revealed a new AI "agent" called Gato that can complete over 600 different tasks throughout a wide range of environments.
This artificial intelligence agent is a single neural network, a computing system with interconnected nodes that work like nerve cells in the human brain.
Gato can chat, caption images, stack blocks with an actual robot arm, and even play Atari, the famous 1980s home video console.
The comments of De Freitas came in reaction to an opinion article published on The Next Web, saying that humans alive today will not attain AGI.
Commenting on the opinion piece, the research scientist tweeted, "It's all about scale now!" He added, 'The Game is Over!" and AI is about making the models larger, safer, computing efficiently, and faster.
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
Freitas admitted, though, that humans are still far from developing an AI that can pass the so-called "Turing test," a test of a machine's ability to display intelligent behavior equivalent to or impalpable from that of a human.
Virtual Assistant
After the announcement of Gato, it is demonstrating AGI, no more than virtual assistants like Alexa of Amazon and Siri of Apple, which already exists on the market and in many people's homes.
The ability of Gato to carry out different tasks is more like a video game console that can store 600 different games compared to a game that can be played in 600 different ways.
Essentially, Gato has been developed to achieve a variety of hundreds of tasks, although the ability may compromise the quality of every task, other commentators said.
Drawbacks
Columnist Tiernan Ray wrote in another opinion article the agent is, in fact, "not so great on several tasks." On the other hand, the program can do better than a dedicated machine learning program when it comes to controlling a robotic "Sawyer arm that stacks blocks," the columnist also said.
The ability of Gato at standard chat dialogue with a human interlocutor is similarly mediocre, sometimes prompting contradictory, not to mention nonsensical utterances. For instance, when a chatbot, Gato, at first, mistakenly said that "Marseille" is France's capital.
Furthermore, a caption Gato created to accompany an image of a red "man holding up a banana to take a picture of it," even though the man was not holding one.
Gato is detailed in a new study posted on the Arxiv preprint server. The firm's authors have said that such an agent will significantly improve performance when scaled up.
Lastly, AGI has already been identified as a future threat that could "wipe out humanity," whether by accident or deliberately.
Related information about DeepMind is shown on AI News's YouTube video below:
RELATED ARTICLE : NVIDIA Omniverse and the Important Role It Plays in All Metaverses
Check out more news and information on Artificial Intelligence in Science Times.
