There’s been widespread chatter around how AI is now approaching the abilities of humans, but Elon Musk has a completely opposite take on the issue.
Elon Musk has hinted that humans might actually be a form of AI but they don’t realize it. “Maybe we are LLMs, living on a GPU & don’t realize it,” Musk tweeted. LLMs stands for Large Language Models, which is the technical name for models such as ChatGPT. GPUs are Graphical Processing Units, the advanced computer chips on which models such as ChatGPT are trained.
It’s a pretty intriguing idea, but this isn’t the first time that Musk has hinted at something along these lines. Musk has previously said that it’s very likely that we’re all living in a simulation. Musk says that the universe is 13.8 billion years old, so any civilizations that may have arisen throughout the cosmos have had loads and loads of time to hone their technological know-how. “If you assume any rate of improvement at all, then games will be indistinguishable from reality, or civilization will end. One of those two things will occur,” Musk says. “Therefore, we are most likely in a simulation, because we exist.”
And the last few months have provided a glimpse of how this might actually be the case. While computer games have been getting more sophisticated over time, with better graphics and more realistic gameplay, the rise of programs such as ChatGPT have completely altered the conception of what machines are capable of. Programs like ChatGPT can write poems and articles, solve captchas, figure out mazes, solve McKinsey case studies, and even draw and paint. It’s not hard to imagine a future where these programs are at par with humans in their abilities, and operate as one of us.
And could there be a future where there are no humans at all, but just AIs interacting with each other? A Google paper recently created such a world — the researchers created 25 “characters” which are language models like GPT, gave them names and personalities, put them in a fictional town, and allowed them to interact with each other. The researchers discovered that these characters interacted with one other, asked each other about their lives, and even threw parties. But what Musk seems to be saying is: How do we know we aren’t in one of these simulations, laughing, talking and throwing parties?
Pingback: GPT-4 Responds To Positive Reinforcement, Is Able To Give Correct Answers When Told It's An Expert - MagicWand AI