xAI, an artificial intelligence startup founded by billionaire entrepreneur Elon Musk, released in early November its first AI chatbot, called Grok. The firm’s CEO claimed in a tweet that “In some important respects, it is the best that currently exists.”
“Grok is an AI modeled after the Hitchhiker’s Guide to the Galaxy, so intended to answer almost anything and, far harder, even suggest what questions to ask! […] It will also answer spicy questions that are rejected by most other AI systems,” the startup said in a statement on its website.
By creating Grok, xAI aims to deliver AI tools that “maximally benefit all of humanity” and “are useful to people of all backgrounds and political views.” Another objective is to empower research and innovation, assisting users in the pursuit of understanding.
The chatbot is also designed to answer questions with a bit of wit and has a rebellious streak, therefore it is not recommended for those who don’t understand humor, the statement reads further.
Grok’s “unique and fundamental advantage […] lies in real-time knowledge of the world via the X platform” [formerly known as Twitter], which Musk mistakenly brands as a social media free of disinformation. In fact, X now is a preferred platform for conspiracy theories, fake news, and propaganda, a reason why many schooled users and advertisers have abandoned it since Musk took control of Twitter in October 2022.
The engine behind Grok is the large language model Grok-1, which has been quietly in the making during the past four months. Grok-1 was trained on data from the open internet and the information provided by AI tutors.
The model can still hallucinate [provide erroneous answers], despite the access to external information sources, xAI admits.
Grok is in the beta version following two months of training, but the team behind the product promises to improve it within weeks with the help of a group of select testers. Musk has recruited for the team specialists who previously worked at DeepMind, OpenAI, Google Research, Microsoft Research, Tesla, and the University of Toronto.
Grok took a B in math
To understand the “intelligence” of its product, xAI assessed Grok’s performance in comparison with OpenAI’s chatbots GPT-3.5 and GPT-4, and Anthropic’s Claude-2 based on the Hungarian national high school finals in mathematics, which was published in May.
Grok passed the exam with a C (59%), ahead of GPT-3.5 (41%) and Claude-2 (55%), but behind GPT-4, which received a B (68%), xAI said.
Elon Musk is not a stranger to AI. He was one of the co-founders of OpenAI, quitting it in 2015 over disagreements with his partners, including Sam Altman, over the development direction. He first publicly admitted to be working on a rival to ChatGPT in April 2023, in an interview on FoxNews. He called the future baby AI "TruthGPT".
At xAI, Musk holds the same position as at Tesla, X, Starlink, and SpaceX: chief executive officer.
Feel free to support our small office: @businessnewsservice (PayPal) or IBAN - RO50BTRLEURCRT0490900501, Swift - BTRLRO22, Beneficiary - Rudeana SRL.
Not feeling like donating? Then click on banners on our website to generate ad revenue. Any help is welcome.