Grok launches touted as ChatGPT killer by Elon Musk-owned xAI
It is pitched as a ChatGPT competitor, with a "unique and fundamental advantage" thanks to the eponymous LLM Grok-1 trained on "current" online information via X/Twitter. xAI claims victory for the model over GPT version 3.5 (but not 4) in a range of benchmarks including MMLU and HumanEval.
Grok-1's grades were also found to come second to those of GPT-4 in Hungary's National High School Math Exam of May 2023), a "real-life", "hand-graded" test that xAI maintains "our model was never explicitly tuned for".
Nevertheless, the company asserts that the rate at which Grok has been developed ensures it is ready to take on chatbot duties with results that outstrip those of GPT in terms of relevance and accuracy.
It is also rated to tackle "spicy questions that are rejected by most other AI systems" in a manner not recommended "if you hate humor(!)" (a phrasing that already hints at just how subjective that apparently inbuilt "wit" might end up getting).
Nevertheless, xAI also claims it is committed to the development of "reliable safeguards against catastrophic forms of malicious use" in a bid to do "our utmost to ensure that AI remains a force for good".
Then again, as it is projected to become restricted to X Premium Plus subscribers following its roll-out, its effects might be limited to those who feel compelled to pay Elon Musk for stuff that used to be free in any case.