Answers Are Not Always Correct
Another drawback is that because it is programmed to give responses that feel natural to people, the answers may lead people to believe that the output is accurate.
Many users observed that ChatGPT sometimes gives false information, including those that are radically false.
It’s possible that responses that seem reasonable to people have an unintended effect, as was observed by the moderators at the coding Q&A website Stack Overflow.
Stack Overflow was overwhelmed with user responses coming from ChatGPT that seemed to be the right answers, but there were actually a lot of them.
The thousands of answers overwhelmed the volunteer moderator team, prompting the administrators to enact a ban against any users who post answers generated from ChatGPT.
The flood of ChatGPT answers resulted in a post entitled: Temporary policy: ChatGPT is banned:
“This is a temporary policy intended to slow down the influx of answers and other content created with ChatGPT.
…The primary problem is that while the answers which ChatGPT produces have a high rate of being incorrect, they typically “look like” they “might” be good…”
The experience of Stack Overflow moderators with wrong ChatGPT answers that look right is something that OpenAI, the makers of ChatGPT, are aware of and warned about in their announcement of the new technology.
OpenAI Explains Limitations of ChatGPT
The OpenAI announcement offered this caveat:
“ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers.
Fixing this issue is challenging, as:
(1) during RL training, there’s currently no source of truth;
(2) training the model to be more cautious causes it to decline questions that it can answer correctly; and
(3) supervised training misleads the model because the ideal answer depends on what the model knows, rather than what the human demonstrator knows.”
Is ChatGPT Free To Use?
Currently, during the “research preview” period, ChatGPT usage is free.
Users can currently test out the chatbot and give feedback on the responses so that the AI can improve at responding to inquiries and learn from its errors.
According to the official statement, OpenAI is happy to hear input regarding the errors:
“While we’ve made efforts to make the model refuse inappropriate requests, it will sometimes respond to harmful instructions or exhibit biased behavior.
We’re using the Moderation API to warn or block certain types of unsafe content, but we expect it to have some false negatives and positives for now.
We’re eager to collect user feedback to aid our ongoing work to improve this system.”
There is currently a contest with a prize of $500 in ChatGPT credits to encourage the public to rate the responses.
“Users are encouraged to provide feedback on problematic model outputs through the UI, as well as on false positives/negatives from the external content filter which is also part of the interface.
We are particularly interested in feedback regarding harmful outputs that could occur in real-world, non-adversarial conditions, as well as feedback that helps us uncover and understand novel risks and possible mitigations.
You can choose to enter the ChatGPT Feedback Contest3 for a chance to win up to $500 in API credits.
Entries can be submitted via the feedback form that is linked in the ChatGPT interface.”
The currently ongoing contest ends at 11:59 p.m. PST on December 31, 2022.
Will Language Models Replace Google Search?
LaMDA is an AI chatbot that Google has already developed. An engineer at Google asserted that LaMDA was sentient since the performance of the chatbot was so similar to a human discussion.
Is it unlikely that a business like OpenAI, Google, or Microsoft will eventually replace conventional search with an AI chatbot given how these massive language models can respond to so many queries?
On Twitter, some people have already predicted that ChatGPT will overtake Google.
For those who make their job as search marketing experts, the possibility that a question-and-answer chatbot would eventually replace Google is terrifying.
It has spurred debates in online communities for search marketing, such as the well-known Facebook SEOSignals Lab, where someone questioned whether or not search queries may shift away from search engines and toward chatbots.
After using ChatGPT, I have to admit that the worry over chatbots taking the role of search engines is not unwarranted.
Although there is a long way to go in terms of technology, it is conceivable to picture a search future that combines chatbots with hybrid search.
But it appears that ChatGPT as it is now implemented will eventually need users to spend credits in order to utilise it.
How Can ChatGPT Be Used?
ChatGPT is capable of crafting text in the form of short stories, poems, songs, and even code.
ChatGPT is transformed from a source of information to a tool that may be used to complete a task thanks to its proficiency in following instructions.
It can therefore be used to write an essay on just about any subject.
ChatGPT can be used as a tool to create article or even book-length outlines.
Almost any assignment that can be answered with written word will have a response from it.
As was already mentioned, it is planned for the public to eventually pay to utilise ChatGPT.
Within the first five days of ChatGPT’s public launch, more than a million users had registered to utilise it.