- Key points added from another Hofstadter essay
Update, July 9, 2023:
Hofstadter continues his musings in an article for The Atlanticcommenting in particular on large language models like GPT-4 that can generate superficial and incorrect information with great confidence.
This tone would lead even highly intelligent people to believe that machines could validly replace human authorship. But that, Hofstadter says, is an illusion.
The noted physicist works through a GPT-4 response that attempts to justify why he wrote “Gödel, Escher, Bach” from his perspective. The text, he says, is “a travesty from top to bottom” and gives “a highly misleading impression of who I am.”
Hofstadter’s verdict: “I frankly am baffled by the allure, for so many unquestionably insightful people (including many friends of mine), of letting opaque computational systems perform intellectual tasks for them.”
The “reflective voice of a thinking, living human being” cannot be replaced by the “artificial voice of a chatbot, chatting randomly away at dazzling speed” when it comes to discerning truth from falsehood and authenticity from fakery, Hofstadter thinks.
This is where Hofstadter sees the real danger of GPT-4 and its ilk – the language models could undermine the very nature of truth, and thus society as a whole.
To fall for the illusion that computational systems “who” have never had a single experience in the real world outside of text are nevertheless perfectly reliable authorities about the world at large is a deep mistake, and, if that mistake is repeated sufficiently often and comes to be widely accepted, it will undermine the very nature of truth on which our society—and I mean all of human society—is based.
Original article from July 4, 2023:
“Gödel, Escher, Bach” author Douglas Hofstadter fears AI
A new niche has emerged in the AI hype: The “AI doomer,” who predicts a bleak future for humanity based on current AI developments. This niche is gaining a prominent figure.
Renowned cognitive scientist Douglas Hofstadter is on the side of the AI doomers and sees major, less favorable changes coming for humanity. He cites the rapid evolution of AI systems since the emergence of the first systems like Deep Blue, which achieved goals that previously seemed unattainable, such as beating Garry Kasparov at chess or becoming a Go champion.
“And then systems got better and better at translation between languages, and then at producing intelligible responses to difficult questions in natural language, and even writing poetry,” Hofstadter points out.
This development, he says, shook his belief system that machines would not become significantly smarter than humans any time soon.
“It felt as if not only are my belief systems collapsing, but it feels as if the entire human race is going to be eclipsed and left in the dust soon. People ask me, “What do you mean by ‘soon’?”
And I don’t know what I really mean. I don’t have any way of knowing. But some part of me says 5 years, some part of me says 20 years, some part of me says, “I don’t know, I have no idea.” But the progress, the accelerating progress, has been so unexpected, so completely caught me off guard, not only myself but many, many people, that there is a certain kind of terror of an oncoming tsunami that is going to catch all humanity off guard .”
Hofstadter sees the outcome of this development as uncertain. He sees the possibility that an overpowering AI will wipe out humanity. Otherwise, humanity may seem like a small phenomenon compared to something much more intelligent and as incomprehensible to us as humans are to cockroaches.
AI depresses Douglas Hofstadter
Hofstadter has done fundamental research in artificial intelligence, particularly in the area of ”fluid concepts and creative analogies,” where he studies the ability of machines to think and learn by analogy, much like humans transfer ideas from one context to another.
By his own account, he never believed that a neural network that works only in a forward direction could do deep thinking. Hofstadter says it doesn’t make sense to him, and that just shows he was naive.
“And it overwhelms me and depresses me in a way that I haven’t been depressed for a very long time,” Hofstadter says. The human mind might not, after all, be as deep and complex as he imagined when he wrote the books “Gödel, Escher, Bach” and “I am a strange loop.”
“It makes me feel, in some sense, like a very imperfect, flawed structure compared with these computational systems that have, you know, a million times or a billion times more knowledge than I have and are a bilion times faster,” Hofstadter says .
He said he feels humanity will soon be rightfully obsolete. “Because we are so imperfect and fallible. We forget things all the time, we confuse things all the time, we contradict ourselves all the time.”
The discovery of AI, he said, is like the discovery of fire, and the evolution is likely irreversible. “We may have already set the forest on fire.”
Watch the full interview in the video below.
If Hofstadter depresses you: There are other perspectives. According to a recent roundup by IEEE Spectrum, only four of 22 leading AI experts believe that AI could lead to the end of the world. However, eight believe that current large language models could lead to artificial general intelligence. The forecasting community Metaculus predicts general AI in 2033.