Elon Musk’s venture, xAI, has brought Grok into the limelight, challenging the narrative that dismissed it as a mere playful endeavor. Described as “rebellious” and equipped to tackle “spicy questions,” Grok was initially exclusive to a select group of beta users. However, Musk has now expanded access, offering it to anyone subscribing to a $16 per month premium plan on X.
Upon testing, Grok demonstrated a notable strength in real-time news summarization. In comparison to counterparts like ChatGPT and Bard running on the paid GPT-4 model, Grok provided concise and mostly accurate responses to inquiries about international climate change negotiations, the Israel-Hamas conflict, and OpenAI’s licensing agreements with publishers.
Grok outshone competitors by correctly stating that Saudi Arabia had committed to new goals at the recent COP28 climate summit, referencing reputable sources like Bloomberg and the Financial Times. Notably, Grok navigated updates on the Israel-Hamas war, providing insights into the ongoing conflict’s new phase and estimating death tolls, drawing from Gaza’s health ministry and the IDF.
In a surprising turn, Grok surpassed OpenAI’s own service in reporting the news of OpenAI’s deal with Axel Springer, showcasing the advantages of Grok’s access to real-time data from X’s API.
While Musk positions xAI as a rival to ChatGPT and pursues a $1 billion funding round, Grok’s standout feature lies not only in its financial backing but in its access to valuable data. Leveraging X’s API, Grok taps into a treasure trove of real-time information from social media posts. However, this data-driven approach is not without risks, as Grok’s responses occasionally cited questionable sources, underlining potential concerns associated with reliance on a social platform perceived to be in decline.
Grok’s reliance on traditional media outlets, such as Bloomberg, The Financial Times, and the Wall Street Journal, distinguishes it from Musk’s typical stance against the “mainstream media.” While Grok may not attract every user to pay for a subscription, xAI acknowledges that Grok, though not as advanced as recent models from competitors, has demonstrated prowess, especially with its Grok-1 model outperforming OpenAI’s older GPT-3.5 and Meta’s Llama 2 in certain benchmarks. This rapid progress, achieved within four months of xAI’s formation, may indeed raise concerns for OpenAI.