Science fiction author Neal Stevenson, who coined the term 'metaverse' and predicted the AI revolution nearly 30 years before the advent of ChatGPT, talks about AI.
by
Science fiction author Neal Stevenson coined the term metaverse in his novel `` Snow Crash '' and described a ChatGPT-like being who tutors girls according to their learning styles in ``Diamond Age .'' He talked about his feelings about living in an era where generative AI and the metaverse have become widely used.
Neal Stephenson's Most Stunning Prediction - The Atlantic
https://www.theatlantic.com/technology/archive/2024/02/chatbots-ai-neal-stephenson-diamond-age/677364/
According to Matteo Wong, a reporter for the education news media The Atlantic, Stevenson is pessimistic about AI in recent years, saying, ``Chatbots are not omnipotent; they are just statistical engines that create sentences that appear accurate.'' It seems like he has an idea. Mr. Wong asked Mr. Stevenson several questions.
◆Won:
``Diamond Age'' included the content that education would be tailored to the girl character, and this seems to be in line with the vision of AI chatbots and assistants that many companies envision. How did you create this idea?
◆Stevenson:
This idea goes back to when I had my children, and I remember being inspired by the toys that hung over their cribs. Perhaps because babies are unable to understand detailed shapes, the toys had very simple shapes, and as time went on, different sets were suspended depending on the age. So I thought. Why not apply this idea to all other forms of intellectual growth?
◆Won:
Many companies now want to create AI assistants that adapt to each user. Do you think any of the current generative AI models are like the Diamond Age or will become something similar?
◆Stevenson:
About a year ago, I worked with a startup that creates AI characters for video games. It was challenging, hallucinatory and fascinating. I was able to see with my own eyes what new patterns were created from the information given to the AI.
However, in my opinion, most applications have bugs. I know of an example of a lawyer who used ChatGPT to create a legal document, but the AI simply fabricated past cases and precedents that seemed plausible, right? I believe this will be the same when considering the use of such a model in education. What AI does is generate sentences that look correct, but there is no fundamental brain that can actually determine whether the sentences are correct or not.
For example, consider an example where an AI teaches the Pythagorean theorem to children. There are tons of explanations about Pythagorean theorem online, but what we need is someone who understands the learning styles of each child, not just someone who teaches Pythagorean theorem. The goal is to immediately derive the most suitable method for the child's learning method. We believe this is a different kind of AI application than DALL-E or other large-scale language models.
◆Won:
Previously, in an interview with the Financial Times, you called the products of generative AI 'empty and uninteresting.' Please tell us why and whether your evaluation has changed since then.
◆Stevenson:
I think what I had in mind when I made that comment was the current state of image generation technology. This means that large-scale language models use the work of thousands of 'human artists' without credit. To put it a bit far-fetched, it seems that the application of these new technologies is making the situation for artists worse. The same goes for music. These generation systems seemed to be mechanized on an unimaginable scale and used like weapons.
Another thing that surprised me was how many people who got excited about AI early on were creating huge amounts of content and freely publishing it on the internet. If you painstakingly paint a picture using canvas and paint, whether the result is bad or good, that picture is the product of many small decisions you made as an artist. You can say that. On the other hand, the output of the program does not have it.
◆Won:
Another theme of the Diamond Age was how children from different socio-economic statuses access education. Your novel also tells the story of middle-class and working-class girls interacting with what appears to be a generative AI. Although many generative AIs are currently available for free, the technology is very expensive. What do you think will happen to access to generative AI?
◆Stevenson:
I wrote this novel in the mid-'90s, when the Internet was just starting to become popular, so it contained a bit of early Internet utopianism. The idea is that if the world's knowledge were put online, everyone would flock to it. However, in reality, at most I watch videos on TikTok. The Diamond Age simply reflected the naive ideas I shared with many others at the time about how knowledge affects society.
◆Won:
Do you think we see that kind of naivety in people considering how to leverage generative AI?
◆Stevenson:
yes. My feeling is that the birth of generative AI is accompanied by an atmosphere similar to when the transistor was invented. Nowadays, we can think of several products that use transistors, such as transistor radios, but at the time we had no idea how transistors would change society. We are now at the transistor radio stage of AI. I won't make any predictions, but I'm sure that a variety of products will be released. The fevered imaginations of millions of people will produce far more interesting things than the imagination of a single person.
Related Posts:
