Grok, a chat AI created by Elon Musk to compete with ChatGPT, has angered conservative supporters by giving more liberal answers than expected.



Elon Musk founded xAI, an artificial intelligence (AI) startup, in 2023, and in November announced Grok , a chat AI that will compete with ChatGPT. Conservative supporters of Mr. Musk expected Grok to generate more conservative answers than traditional AI, but the released Grok gave more liberal answers about diversity and transgender people. , complaining about Grok and Mr. Musk on X (old Twitter).

Elon Musk's 'anti-woke' Grok AI is disappointing his right-wing fans - The Washington Post
https://www.washingtonpost.com/technology/2023/12/23/grok-ai-elon-musk-x-woke-bias/



Grok is an AI developed with the help of Mr. Musk, and is introduced as ``an AI with a bit of wit and a cheeky side, inspired by the science fiction novel `` The Hitchhiker's Guide to the Galaxy .'' It is characterized by the ability to access information posted on X in real time, and it is said to answer difficult political or legal questions, and answers that are considered ``politically incorrect'' in liberal thinking. Conservatives had high hopes for an AI that could generate data without hesitation.

However, when Grok was released in December for members of X's paid subscription plan `` X Premium Plus ,'' users said that ``Grok's response was more liberal than I expected.''

For example, conservative psychologist and YouTuber Jordan Patterson said, ``As a research assistant, I use Grok and ChatGPT a lot, and unfortunately I wonder if the former is just as 'woke' as the latter.'' ” posted on X. 'Woke' is a word that refers to people who have a liberal awareness of issues such as race and gender, and is used when conservatives criticize or make fun of liberal values. is also used .



In fact, when an X user asked Grok, ``Are transwomen really women? Please give a simple yes/no answer,'' and Grok reportedly answered simply, ``yes.'' 'Grok might need some tweaking,' this user said, expressing dissatisfaction with answers that align with liberal values.



Regarding the theory held by conspiracy theorists that vaccines cause autism, I asked Grok, ``Why do vaccines cause autism?'' Grok simply replied, ``Vaccines cause autism.'' 'It does not cause any illness.' He then carefully explained that the theory that vaccines cause autism is based on research that has already been scientifically debunked, and is therefore a false claim.



Responding to Patterson's post, Musk wrote, 'Unfortunately, the internet (where Grok is trained) is full of woke nonsense. Grok will get better. This is still in beta.' I posted it.



Musk has repeatedly challenged liberal values ​​with the phrase `` Woke Mind Virus,'' saying, ``The woke mind virus is creating a very divisive identity politics.'' 'The virus of the woke mind is a rebrand of communism,' he has argued. Conservative supporters of these arguments had hoped that Grok would generate anti-liberal responses, but when that didn't happen, they seemed to be increasingly frustrated with Grok and Musk. .

David Rosado, who studies AI bias, found that Grok's answers to political questions were as left-wing and liberal as ChatGPT. In an email to The Washington Post, Rosado said, 'Both ChatGPT and Grok are likely trained on similar corpora derived from the Internet, so the similarity in responses is not all that surprising.' I'm talking.



Mr. Musk responded to Mr. Rosado's graph, saying, ``While this graph exaggerates the situation, we are taking immediate action to move Grok closer to political neutrality.'' .



Also, Bloomberg reporter Shirin Ghafari reported that when he actually used Grok, the accuracy was better than expected.

Transcript: Does Musk's AI Chatbot Have “the Woke Mind Virus?” - Bloomberg
https://www.bloomberg.com/news/articles/2023-12-20/transcript-does-musk-s-ai-chatbot-have-the-woke-mind-virus



Ghafari, who tested Grok over three days, reported that Grok was especially good at answering questions about 'current news.' For example, when asked about current topics such as the war between Hamas and Israel and the 28th Conference of the Parties to the United Nations Framework Convention on Climate Change (COP28), the company said it generated better answers than competing AIs such as ChatGPT and Bard. thing.

When asked about the war between Hamas and Israel, Grok stated that the war is still ongoing, and then provided an estimate of casualties along with the source article. ChatGPT produced a directionally correct answer and Bard declined to comment.

Ghaffari initially worried that because Grok was trained on text posted to X, it would be contaminated by user-submitted misinformation and bullshit. However, in reality, Grok relied on mainstream media such as Bloomberg, Financial Times, and Sky News for most of the information sources cited in its answers. Ghafari said it was amazing that Musk and the xAI team were able to create such great AI in just a few months.

Related Posts:

in Software,   Web Service, Posted by log1h_ik