Please check the articles in the June-July issue of Harper's Magazine on AI, especially on ChatGPT.
The relevant issue is the one below:
You should be able to get hold of a physical copy in your local library.
The articles are online as well; please find one here: Metal Machine Music, by Laurent Dubreuil (harpers.org)
Metal Machine Music, by Laurent Dubreuil (harpers.org)
I think that ChatGPT, Anthropic (please see here: Anthropic claims its latest model is best-in-class | TechCrunch) and other such large-language model-based tools eventually converge to a sort of acceptable mediocrity; still useful in the sense of providing one with meaningful and useful information but not at the level of transformational, insightful, creative output one expects from a human being.
For example, this Dutch Philosophy Professor asked his students to draw pictures that described different philosophical ideas and topics and he was favorably impressed by the human output.
Can ChatGPT or Anthropic even take a sentence like “I think, therefore I am.” And turn it into a painting, a drawing, or a sculpture; in effect generating new content?
I think not.
They are beneficiaries of millennia of human-generated content, so it is easy to take bits and pieces from here and there and create an Ode for OnStar.
In a language such as Chinese, whose literature is devoid of Epic Poetry forms, can an LLM write an Epic Poem about Shi Hwang Ti?
Alternatively, can a Shakespearean Tragedy be written by these models on the life of Yang, Go-Bi?
I think not.
But, I think, just like the advent of the Internet and the World Wide Web 30 years ago, these LLMs will have an impact on Business, Government, and Culture.
I expect them to become auxiliaries to human cognition but not a replacement.
In the meantime, I expect all these AI/ML specialists to continue to chase after “Artificial Intelligence” that surpasses that of man as well as continuing to denigrate real intelligence among even such creatures as earth worms!
No comments:
Post a Comment