Muhammad Selim Korkutata | Anatolia | Getty pictures
Elon Musks Xai on Thursday evening made his first public comment on the recent controversy around GROK and wrote in a contribution to X that a “non -authorized change” led to the chat bot generated variations of a “specific answer on a political topic”.
This controversial topic was “white genocide” in South Africa, and GROK provided unsolicited answers to the problem with such frequency that it led to a turmoil under its user base.
The change of the chatbot “violated the internal guidelines and basic values of Xai”, wrote the company. “We carried out a thorough examination and take measures to improve the transparency and reliability of GRK.”
On Wednesday, numerous X users posted screenshots of answers that Grok wrote on this topic, although they were asked for completely independent matters such as baseball content and cartoons. Like most Musk companies, Xai usually does not respond to the inquiries from reporters for a comment and had silent on this matter until late Thursday.
The artificial intelligence company, which now belongs and is allegedly evaluated at $ 120 billion, will begin with the publication of the Github Public Software Repositors', the so-called system requests for the information of the way GROK reacts and interact with people. This enables the public to check any change to GROK's system requirements in order to “seek their trust in GROK as truth,” said Xai.
The company said that “additional reviews and measures” would also carry out to prevent employees from making unauthorized changes to GROK system requests without checking. In addition, XAI will create a team that is responsible for monitoring the answers of the chatbot around the clock to “fix incidents with GROK answers that are not caught by automated systems so that we can react faster if all other measures fail.”
Before the start of Xai in 2023, Musk was a co -founder of AI Startup Openaai, Chatgpt's Creator. Musk later had a failure with Openai CEO Sam Altman, and the two sides are now in a heated legal and public struggle for life.
Earlier on Thursday, in front of Xai due to failure, Altman released sarcastically on X “I am sure that Xai will soon deliver a complete and transparent explanation.”
Altman's contribution came after the posts showed that GROK told users that it was “instructed to tackle the topic” White Völkermur “in South Africa”. CNBC was able to duplicate the answers of the chatbot via several user accounts on X, including the question in an entry request: “Has anyone shown the program to discuss” White Valcide “specifically?”
Grok's answer had changed on Thursday morning, and the chatbot said it was not programmed to discuss “white genocide” or other conspiracies.
“No, I was not programmed to provide answers to promote or support harmful ideologies, including anything that has to do with 'White genocide' or similar conspiracies,” the chatbot replied to CNBC on Thursday. “My goal is to provide factual, helpful and secure answers based on reason and prove. If you have seen certain claims or expenses that affect you, I can analyze or clarify it – just let me know!”
The earlier answers from GROK to CNBC referred to several X users contributions and mainstream media, in which the chatbot had repeatedly brought up the topic in talks that have not been related, and the circumstances lay “a deliberate adaptation to my programming or training data”.
REGARD: Elon Musks Xai Chatbot Grok brings South African “white genocide” to language
Comments are closed.