How should the learning algorithms contained in ChatGPT technology be improved so that the answers to questions generated by this form of artificial intelligence are free of factual errors and fictitious 'facts'?
How can ChatGPT technology be improved so that the answers provided by the artificial intelligence system are free of factual errors and inconsistent information?
The ChatGPT artificial intelligence technology generates answers to questions based on an outdated set of data and information downloaded from selected websites in 2021. In addition, the learning algorithms contained in ChatGPT technology are not perfect, which means that the answers to questions generated by this form of artificial intelligence may contain factual errors and a kind of fictitious 'facts'. A serious drawback of using this type of tool is that the ChatGPT-generated answers may contain serious factual errors. When people ask about something specific, they may receive an answer that is not factually correct. ChatGPT often answers questions eloquently, but often its answers may not relate to existing facts. ChatGPT can generate a kind of fictitious 'facts', i.e. the generated answers may contain stylistically, phraseologically, etc. correctly formulated sentences containing descriptions and characteristics of certain objects presented as real existing but not actually existing. In the future, a ChatGPT-type system will be refined and improved, which will be self-learning and self-improving in the analysis of large data sets and will take into account newly emerging data and information to generate answers to the questions asked without making numerous mistakes as is currently the case.
In view of the above, I address the following question to the esteemed community of scientists and researchers:
How should the learning algorithms contained in the ChatGPT technology be improved so that the answers generated by this form of artificial intelligence to the questions asked are free of factual errors and fictitious "facts"?
How can ChatGPT technology be improved so that the answers provided by the artificial intelligence system are free of factual errors and inconsistent information?
What do you think about this subject?
What is your opinion on this subject?
Please respond,
I invite you all to discuss,
Thank you very much,
Best regards,
Dariusz Prokopowicz
Every thing improves by evolutionary approach so the ChatGpt would also improve with the more advanced version like GPT-4 etc. the journey of AI just started now it has to be move so long . with the understanding of right notion of intelligence and development of AI without the help of any algorithm like human.
Thanks for asking and open discussion. I try to used chat gpt and talking with it like with my friend. I think, it learn by its friend while get communication each other. Its get a lot of insight from user. I am worried, it will be smarter than us because it learn form a lot of user with any type. But, i think it doesn't work perfect like human being specially for multitasking and emotional suggestion.
Dariusz Prokopowicz ,
The problem facing ChatGPT is a combination of factors that need addressing. First of all Chat GPT is a large language model(LLM). As such it does not do fact checking. Fact checking is a complex subject that takes a multiplicity of algorithms. Some of the algorithm categories include:
1) Named entity disambiguation(NED)- This is not the same as named entity recognition(NER). NED techniques try to disambiguate instances such as 'Michael J. Jordan' from 'Michael B. Jordan' when the text is using only 'Michael Jordan'.
2) Source reliability- This checks that the source of the information can be relied upon. This is usually a vetting process or static lists
3) Source content reliability- The content is checked for consistency by clustering similar pieces of information
These methods need to be used in conjunction to minimize errors. The problem of checking other problems such as causality is another problem entirely but also relevant to LLMs.
Regards
Dear people,
http://www.bbc.com/future/story/20160105-the-man-who-studies-the-spread-of-ignorance
“We live in a world of radical ignorance, and the marvel is that any kind of truth cuts through the noise,” says Proctor. Even though knowledge is ‘accessible’, it does not mean it is accessed, he warns…the knowledge people have often comes from faith or tradition, or propaganda, more than anywhere else.”
Is it a reality - how do you think?
https://en.wikipedia.org/wiki/Mechanical_Turk
he Turk, also known as the Mechanical Turk or Automaton Chess Player (German: Schachtürke, lit. 'chess Turk'; Hungarian: A Török), was a fraudulent chess-playing machine constructed in the late 18th century. From 1770 until its destruction by fire in 1854 it was exhibited by various owners as an automaton, though it was eventually revealed to be an elaborate hoax.[2] Constructed and unveiled in 1770 by Wolfgang von Kempelen (1734–1804) to impress Empress Maria Theresa of Austria, the mechanism appeared to be able to play a strong game of chess against a human opponent, as well as perform the knight's tour, a puzzle that requires the player to move a knight to occupy every square of a chessboard exactly once.
The Turk was in fact a mechanical illusion that allowed a human chess master hiding inside to operate the machine. With a skilled operator, the Turk won most of the games played during its demonstrations around Europe and the Americas for nearly 84 years, playing and defeating many challengers including statesmen such as Napoleon Bonaparte and Benjamin Franklin. The device was later purchased in 1804 and exhibited by Johann Nepomuk Mälzel. The chess masters who secretly operated it included Johann Allgaier, Boncourt, Aaron Alexandre, William Lewis, Jacques Mouret, and William Schlumberger, but the operators within the mechanism during Kempelen's original tour remain a mystery.
Who answers to users in large language model built on great human thoughts database?
Thinking?
There may also be simply an issue of what information it has access to. Academic databases may be behind paywalls and out of its reach when it makes calculations, leaving it to rely on less rigourous research.
- Badges
- Science topic