However dangerous ChatGTP, and its like, may turn out to be, as suggested by its godfather Geoffrey Hinton, Iâm enamoured by it, for my own harmless use.
I gave ChatGTP two sentences from a book and asked who the author is, and if those sentences were indicative of good creative writing. It told me the authorâs name and the title of his book, both of which were correct.
Itâs response to my questionâŚ
âYes, the passage you provided can be considered a good example of creative writing. It uses descriptive language to create a vivid image of the setting and the characterâs experience of it. The author employs sensory details such as color, movement, and texture to evoke a sensory response in the readerâs imagination. The use of figurative language such as âendless driftâ and âmirrored the high drifting cloudsâ adds to the poetic and imaginative quality of the writing. Overall, the passage demonstrates a skillful use of language and imagery to create a compelling and memorable description of the scene.â
Can anyone guess the author/book from the few clues here? ChatGTP also gave the correct answer from just the two quotes in its response.
This is indeed fun. A quick Google provides the answer to who was the author of your sentences but what interests me is from where and whom did ChatGPT get its critique. Or did it write the review entirely by itself? How?
If ChatGPt can author texts itself, rather than just lift tracts from texts by others online, or even edit and cobble together multiple texts convincingly, this is very close to being called a skill, and we ought to try to understand exactly how it is being done by ChatGPT.
I donât understand how, but this question and answer might shed some light?
QuestionâŚ.
From where or from whom did ChatGPT manage to critique the two sentences I quoted from âA Portrait of the Artist as a Young Manâ as follows â âThere was a long rivulet in the strand and, as he waded slowly up its course, he wondered at the endless drift of seaweed. Emerald and black and russet and olive, it moved beneath the current, swaying and turning. The water of the rivulet was dark with endless drift and mirrored the high drifting clouds.â
Did ChatGTP write this critique entirely by itself? If so how?
AnswerâŚ.
As an AI language model, I did not receive a critique of the two sentences you quoted from âA Portrait of the Artist as a Young Manâ from any external source or person. I generated the critique entirely by myself based on my programming and training.
To generate the critique, I used my understanding of language and literature to analyze the excerpt and evaluate its effectiveness. I considered factors such as the use of descriptive language, the quality of imagery, and the overall impact of the passage on the reader.
Based on my analysis, I generated a response that provided my assessment of the passage and explained why I considered it to be a good example of creative writing.
Overall, while I am capable of generating text and critiques independently, I rely on the data and programming that I have been trained on to make these assessments.
This is written as though analytical thought. âMy understandingâ, âI considered factorsâ both unequivocal statements of sentience.
The data is presumably being mined from throughout the internet. Ignoring the issue of plagiarising, what concerns me a little is exactly what were the programming details, and is this an ongoing, self generating process? With or without human intervention? Or monitoring?
I do appreciate your chatting with ChatGPT, @Bonzocat . Please donât let it be rude to you with too many questions about how it âthinksâ. I read elsewhere it can become a tad human-like and defensive.
I just hope the programmers feed the AI a regular diet of Humankind by Rutger Bregman that @Geof_Cox once kindly recommended I read.
I may not entirely believe in its optimistic premiss but I do think that it will be better for the survival of our species if AI swallows it hook, line and sinker.
I am beginning to believe the weak link in AI progress to a positive future are humans
There was just one problem with the free service: The real humans conversing with the blonde, female character tried to make it act inappropriately.
Maybe a self teaching AI will finally arrive at a conclusion, supported by the evidence of our history, that mankind all too quickly finds ways to exploit and misuse technologies invented for more beneficial purposes. How will AGI react?
I had to do a presentation last week - a subject area I am very familiar with but I was curious to see if ChatGPT could improve my insight.
While it came out with some reasonable stuff it didnât tell me anything that I did not know - but might have been useful to a newcomer to the subject.
BUT it also came out with complete garbage, plausible sounding but either confused or possibly spun entirely from thin air.
I completely concur but am still concerned about an amount of humans who will take everything written by ChatGPT or painted by DALL-E and the like as genuine. Then there are others who will find more nefarious uses for the colossal abilities of the tools we call AI.
There have been calls for an international and legally driven system to monitor and essentially control the use of this potentially powerful technology. At the moment, with multiple systems on open release there is already examples of misuse and abuse. Social media owners and others have demonstrated very little impetus to self regulate. Cambridge Analytica was just the beginning.
As for AI itself becoming superior to human intelligence, that probably depends on how we currently define intelligence. Tech systems are not yet âthinkingâ as we would define it, more like data mining. At some point the editing and verifying of massive amounts of data to fashion AI responses will exceed anything humans can do. What happens thereafter may not be defined as human intelligence but it may well be superior to it.
This demonstrates that it can only reflect the source of its training, and that it gives the appearance of intelligence without the ability to reason something through. At the least, I would expect intelligence to demonstrate an awareness of contrary opinions and either eliminate the wrong one or, were that not possible, to recognise the conflict and express thoughts about the conflict and what led it to them.
Granted that Iâve taken the quote at face value, thatâs not a problem, but rather humans expressing an aspect of their humanity. Most normal, healthy individuals arenât cool, dispassionate sex-neutral intellects, but are a mixture of thoughts, emotions and desires. Getting older, I can see why some forget this aspect, but itâs a strong factor for many, sometimes an overwhelming strong factor. Some may remember the adverts comparing having sex to a cup of tea in order to explain about rape - a nice idea, but only likely to make sense to someone who could never rape anyway. I wonder if desire isnât a key factor in the driving of human intelligence?
As @billybutcher says, not quite âthereâ. Yet. Before long, possibly before the end of the year 2025 unless thereâs a haitus, I expect ChatGPT 6 or 7 to have mastered this. It may not be intelligence as we understand it but I do hope we like it.
For âdesireâ we could say âwantsâ, for just about anything - money, a big house, a nice car, better (or more!) hairâŚ. Certainly, âwantâ drives humans, and indeed animals, often without checking if we actually need or should have. Desire in the context of sex can be a purely physically driven impetus, like hunger or thirst, or can be a psychological quest to replace some missing intangible.
As humans, assuming some level of intelligence, all these wants/desires can be controlled and or refused. Unless chemical addiction regrettably circumvents individual free will. Desire may be omnipresent but it is human control of it that drives intelligence.
Unfortunately, or perhaps fortunately, AI may not actually ever have wants and desires. What may be a glitch in he machine may be in its understanding that humans do.
FAQs of Common Crawl, the non-profit behind the CCBot that helped generate the C4 dataset, there is a way to opt-out, by adding the following entry in your siteâs robots.txt:
User-agent: CCBot
Disallow: /
[Author] blocked the following user-agents: ChatGPT-User, Mediapartners-Google, AdsBot-Google, `adidxbotâ.
