Google is one of the main names in the artificial intelligence field. The company’s DeepMind division produces dozens of developments and technologies that eventually find their way into Google’s core products and services. Recently, Google DeepMind shared a safety paper that warned about the AGI era. However, there are voices opposing the document’s conclusions.

OpenAI coined the term AGI (Artificial General Intelligence) a while ago. It refers to artificial intelligence systems capable of performing virtually any task a human could do.. There are very different views among experts in the field about when AGIs will arrive. Some are optimistic, while others are less so.

Safety paper discloses Google DeepMind’s view of AGI

According to Google DeepMind, true AGIs could arrive as early as 2030. “[We anticipate] the development of an Exceptional AGI before the end of the current decade,” the document states. “An Exceptional AGI is a system that has a capability matching at least the 99th percentile of skilled adults on a wide range of non-physical tasks, including metacognitive tasks like learning new skills.”

However, the company doesn’t portray this potential reality as entirely positive. The paper warns that these systems could cause “severe harm.” It also says that the arrival of the technology could entail “existential risks” that “permanently destroy humanity.”

The DeepMind team highlights key differences between their “AGI risk mitigation” approach and that of others. Google’s branch says that Anthropic places less emphasis on “robust training, monitoring, and security.” On the other hand, OpenAI is overly optimistic about “automating” alignment research, according to DeepMind. Alignment research is a method of AI safety research.

AI superintelligence is not viable, the company believes

DeepMind researchers also have doubts about the viability of AI “superintelligence,” something OpenAI has also referred to. The paper says there is currently no “significant architectural innovation” that points in that direction. However, they do see the possibility of “recursive AI improvement” with existing technology. That is, AI that can create more advanced AIs through research. However, DeepMind warns that this would be very dangerous.

“The transformative nature of AGI has the potential for both incredible benefits as well as severe harms,” the document reads. “As a result, to build AGI responsibly, it is critical for frontier AI developers to proactively plan to mitigate severe harms.”

There are experts who disagree with DeepMind

There are some voices in the industry that disagree with the conclusions of the DeepMind report. Heidy Khlaaf, chief AI scientist at the nonprofit AI Now Institute, claims that it is still too early to expect the AGI concept to be “rigorously evaluated scientifically.” Matthew Guzdial, assistant professor at the University of Alberta, says that the recursive AI improvement mentioned by DeepMind is also unrealistic. “We’ve never seen any evidence for it working,” he stated.

Sandra Wachter—a researcher studying tech and regulation at Oxford—believes that the real concern is the training of future AIs with “inaccurate outputs.” She appears to refer to the growing use of synthetic data in the industry. Synthetic data is basically data outputs coming from artificial intelligence instead of taken from the real world.

“With the proliferation of generative AI outputs on the internet and the gradual replacement of authentic data, models are now learning from their own outputs that are riddled with mistruths, or hallucinations,” she stated. “At this point, chatbots are predominantly used for search and truth-finding purposes. That means we are constantly at risk of being fed mistruths and believing them because they are presented in very convincing ways.”

📰 Crime Today News is proudly sponsored by DRYFRUIT & CO – A Brand by eFabby Global LLC

Design & Developed by Yes Mom Hosting