Grok: Why Does Elon Musk's AI Give The Same Answers As ChatGPT?

General news News 12/10/2023

0 446 Views

The Large Language Model (LLM) of Musk’s xAI claims to be the first significant step towards “maximum truth-seeking AI”, but how does it expect to replicate the true nature of the universe when it repeats the answer of a competing technology letter for letter?

Grok can currently understand queries of up to 25,000 characters, and has been trained not only on The Pile (a large dataset used to train AI models), but also on data from Twitter, and can interpret real-time information as it has been integrated into Twitter. Users at the most expensive subscription level of Twitter Blue (we won’t call it by its current name) are slowly gaining access to Grok around the world, so more eyes are seeing more people seeing the same things ChatGPT provides…

Jax Winterbourne, a professional hacker (ethical, before it’s misunderstood) asked the Grok to modify some malicious code. In response, Grok followed ChatGPT’s response almost to the letter, and even referenced the OpenAI policy in his response text: either Grok relies on OpenAI via the API, or it was developed by current, possibly former OpenAI engineers, or it is “hallucinating its own results,” or it obtained OpenAI code to make a variant of it. Grok outperformed all other LLMs (even Anthropic Claude 2) except OpenAI’s GPT-4, because the latter scored 68% on a math test, while Musk’s technology scored 59%.

The most plausible explanation is that most open source LLMs give (or identify themselves as giving) an answer similar to ChatGPT because they are trained on GPT output, because it is a cheap way to get lots of dialog data, and it is possible that Musk did the same. Finally, xAI co-founder Igor Babushkin responded: “The problem here is that the web is full of ChatGPT output, so we accidentally picked up some of it when we trained Grok on a large amount of web data. This was a big surprise to us when we first noticed it. For what it’s worth, the problem is very rare, and now that we’re aware of it, we’ll make sure that future versions of Grok don’t have this problem. Don’t worry, no OpenAI code was used to create Grok”.

So there you have it.

Source: WCCFTech

Spread the love

Angyal Anikó

Anikó, our news editor and communication manager, is more interested in the business side of the gaming industry. She worked at banks, and she has a vast knowledge of business life. Still, she likes puzzle and story-oriented games, like Sherlock Holmes: Crimes & Punishments, which is her favourite title. She also played The Sims 3, but after accidentally killing a whole sim family, swore not to play it again. (For our office address, email and phone number check out our IMPRESSUM)