Grok AI’s unexpected mirroring of ChatGPT responses sparks debate on AI originality.
Elon Musk’s xAI has touted its Grok Large Language Model (LLM) as the first big step toward creating a maximum truth-seeking AI, one that understands the fundamental essence of the cosmos. However, for the time being, the AI model appears content to repeat exactly the responses of OpenAI’s GPT LLM, a sharp contrast from the overarching aspirational ideals that allegedly define the core ethos of Grok AI.
Grok can currently comprehend a prompt of up to 25,000 characters, per WCCFTECH. The LLM was trained not just on The Pile, a common AI model training data set, but also on mounds of data collected from X. Furthermore, Grok appears to be able to access and evaluate real-time data through its interaction with the X social networking platform.
Elon Musk revealed this week that the Grok AI model is now available to all paying X platform subscribers. Jax Winterbourne, a professional hacker, urged Grok to change dangerous code in order to test this new paradigm. In response, the LLM repeated OpenAI’s GPT response word for word, even referencing OpenAI’s policy in the result language.
Winterbourne then proposes several explanations for such obvious regurgitation, ranging from the cheeky notion that Grok is simply a derivation of OpenAI’s GPT LLM to the far more sensible explanation that the regurgitated response is the consequence of model delusion.
On a held-out math exam, Grok outscored every other LLM, including Anthropic’s Claude 2, with the exception of OpenAI’s GPT-4, earning a total score of 59 percent vs. 68 percent for GPT-4. This shows that the AI model is more than just a variation on OpenAI’s GPT LLM.
As a result, the most likely explanation for this behavior is that Grok has been intensively trained on GPT’s replies. As a result, rather than developing a unique response while referencing xAI’s regulations on malicious code, the LLM merely regurgitated OpenAI’s position. This also demonstrates that today’s AI models are essentially glorified versions of a Chinese room — a thought experiment positing that AI models don’t genuinely understand language or think.
Meanwhile, @DeFiIgnas, a pseudonymous DeFi analyst, has released a detailed update on Elon Musk’s Grok AI, recognizing its ability to boost crypto study. The AI-powered tool has shown expertise in discovering new airdrops, explaining difficult protocol functions, and even creating amusing social media roasts based on user posts.
The analyst, per Coingape, discussed the findings of his studies with Grok, highlighting both its successes and flaws. Grok effectively highlighted the key milestones of the Frax (FRAX) stablecoin roadmap, demonstrating its ability to deliver in-depth insights into individual projects.
Grok AI has demonstrated its creative side by creating a stand-up screenplay roasting social media posts by @DeFiIgnas. This demonstrates Grok AI’s adaptability in interpreting and evaluating various forms of information in the crypto world.
According to a previous report, Grok AI has stepped into the domain of price forecasts, examining current Bitcoin price behavior and anticipating a possible comeback to its all-time high in 2024. While the forecasts were vague, the AI ascribed the expected spike to rising use, Bitcoin ordinals, and institutional investments.
Despite its benefits, Grok AI has some limits. @DeFiIgnas complained about the tool’s failure to identify trending tokens, noting issues with irrelevant information and erroneous listings. The analyst also pointed out errors in Grok’s identification of probable crypto airdrops, such as centralized exchange campaigns that do not compare to retroactive airdrops.
Furthermore, despite the fact that both the Arbitrum (ARB) and Optimism (OP) airdrops had already concluded, Grok AI labeled them as potential distributions. These mistakes highlight the difficulties and flaws of AI-powered crypto research tools.
In conclusion, Elon Musk’s Grok AI, while claiming to be a groundbreaking truth-seeking AI, is currently mirroring the responses of OpenAI’s GPT LLM. While it has shown promise in the field of crypto research, there are limitations and concerns regarding its originality and ability to think independently. As AI continues to evolve, it is crucial to question the extent of its true understanding and creativity.
