A prominent figure, recognized as a key architect of the artificial intelligence (AI) field, has resigned from his position, issuing cautionary remarks about the increasing hazards of AI advancements.
In a statement given to the New York Times, Geoffrey Hinton, 75, declared his resignation from Google, expressing remorse for his contributions. He shared with the BBC his concerns regarding the potentially alarming risks of AI chatbots, stating that although they presently do not exhibit greater intelligence than humans, he believes that they will soon surpass us in this regard.
Dr. Hinton's groundbreaking work on deep learning and neural networks has paved the way for modern AI systems such as ChatGPT. However, the British-Canadian cognitive psychologist and computer scientist cautioned the BBC that chatbots may soon exceed the amount of knowledge stored in a human brain.
"At present, we are observing that GPT-4 surpasses an individual in terms of its comprehensive knowledge base, by a considerable margin. Although its reasoning abilities are not yet as advanced, it can already perform elementary reasoning tasks. However, considering the rapid pace of development, we should be concerned about the future," cautioned Hinton.
In the article featured in the New York Times, Dr. Hinton made reference to individuals with malicious intent who may attempt to utilize AI for harmful purposes. When pressed by the BBC to elaborate on this, he provided an example of a hypothetical, worst-case scenario.
"For instance, envision a scenario where a malevolent actor like Vladimir Putin empowers robots with the capability to establish their own sub-objectives," he stated.
Hinton cautioned that this could potentially result in the creation of sub-goals such as "I need to gain more power." He further stated, "I've realized that the kind of intelligence we're cultivating differs greatly from the intelligence we possess. We are biological entities, while these are digital systems. The fundamental contrast lies in the fact that digital systems contain numerous replicas of the same weights and model of the world."
Dr. Hinton explained that all of these replicas can learn individually but instantaneously share their knowledge with one another. It's akin to having 10,000 people where, whenever one person learns something new, everyone else automatically gains that knowledge. This is how chatbots can accumulate knowledge far beyond what any individual can possess.
Hinton also disclosed that there were several other reasons for his departure from Google. "Firstly, I am 75 years old, so it's time for me to retire. Secondly, I genuinely want to express positive sentiments about Google, and they will be more convincing if I am not employed by them."
Hinton emphasized that he did not wish to denigrate Google and acknowledged that the technology behemoth had been "extremely conscientious."
Jeff Dean, Google's chief scientist, affirmed the company's pledge to approach AI in a responsible manner, stating in a statement, "We remain resolute in our commitment to a responsible AI approach. We are continually advancing our knowledge to recognize potential risks while also pursuing bold innovations."