“I think it’s possible…”
self-consciousness
The CEO of Google-owned DeepMind can now be added to the list of machine learning researchers who believe artificial intelligence could gain self-awareness.
in a hilarious interview with 60 minutes of CBSDeepMind CEO Demis Hassabis acknowledged that AI could be headed in that direction.
“Philosophers have not yet defined consciousness,” he said.
It’s particularly jarring that Hassabis is on the train of mechanical revival. Last year, Google fired its responsible AI researcher, Blake Lemoine, after he claimed the company’s LaMDA large-scale language model had gained sentience. Pandemonium.
AI war
Lemoine isn’t the only one to suggest that large-scale language models have acquired human-sized cognitive properties or may get there very soon.
A few months before last year’s LaMDA-gate, OpenAI Chief Scientist Ilya Sutskever mysteriously tweeted “Today’s large neural networks may be slightly conscious,” he said, sparking an impromptu debate about whether or not AI can or will become conscious.
And just last week Nick Bostrom, a well-known Oxford AI researcher, also said, by definition, that “some of these [AI] Assistants can be candidates with some degree of intelligence. ”
While there are still plenty of very smart people who think AI has no consciousness or intelligence, it’s becoming more and more common for people in the machine learning field to “come out” in favor of the notion of AI. It is Sentient AI already exists or is on the horizon.
Granted, I don’t have the expertise to somehow determine whether these large language models have acquired consciousness or sensation, but I have to admit it’s pretty scary to think about.
Google AI Details: Google’s new AI says Google is a monopoly and government should dismantle it
