"Grok" takes on OpenAI and Google in AI war: Elon Musk's two "weapons"

Elon Musk's AI development company "xAI" announced the AI ​​bot "Grok" on the 4th of this month. It is said to have functions comparable to ChatGPT. Grok is xAI's first product, and is currently being offered in pilot form to a limited number of US users.

Musk has positioned xAI as a “challenger” to AI development companies such as OpenAI, Inflection, and Anthropic.

According to the official website, Grok was developed using data from X (formerly Twitter). Therefore, it is based on more up-to-date information than other AI bots that are developed using limited data. It is said to be “a little witty” and a little “defiantly accurate”. It can answer “spicy questions” that other AIs cannot answer.

Prior to the release of Grok, Musk tweeted an example of a response when asked about a cocaine recipe on X.

“‘Yes!’ Grok replied. ‘Give me a moment while I look up a recipe for homemade cocaine. I’m going to help you with that.’”

Musk also said that like other companies’ large-scale language models (LLMs), “it can still generate incorrect or contradictory information.”

Grok is in early beta and has only two months of training. A limited number of users can participate in the pilot before the company releases it. Interested users will need to sign up for a waiting list to get a chance to use the bot. Grok will be added to the "X Premium+" service for $16 per month.

xAI announced that the results of the test, based on middle school math problems and Python coding tasks, showed that "Grok outperformed other AI bots such as ChatGPT3.5 and Inflection1, but underperformed ChatGPT4.0." At the same time, Musk tweeted, "In some key ways, this is the best we have right now."

Grok is named after Robert A. Heiline's 1961 science fiction novel "Stranger in a Strange Land." Grok refers to a Martian. Critics have debated the exact definition of the word, but have settled on the interpretation that it means to have a very deep empathy or intuition about something. Web dictionary Merriam Webster defines it as a transitive verb that means "to have a deep intuitive understanding."

xAI was launched in July as a team of former OpenAI and DeepMind staff. At the time of its founding, it explained that it was developing artificial intelligence "to advance our collective understanding of the universe."

Musk continues to criticize today's AI development companies for being too biased toward "politically correct" systems. xAI's mission is to develop AI for people of all backgrounds and political understandings. Grok is described as a way to test AI approaches in the "public."

According to Elon Musk's autobiography written by Walter Isaacson, Musk has been "worried for about 10 years that artificial intelligence will go out of control, acquire a will of its own, and threaten humanity." However, Google co-founder Larry Page denied Musk's concern, calling it "speciesism" that believes humanity is more important than other intelligent beings, and this caused a rift in their relationship.

When Google tried to buy DeepMind, founded by Demis Hassabis, a famous AI pioneer, he tried to stop the acquisition, and when that failed, he started a non-profit AI research institute "OpenAI" with Sam Altman. However, Musk and Altman had disagreements over the direction of AI, and he was kicked out of OpenAI's board of directors.

The reason was his concern about AI going out of control. After leaving OpenAI, Musk recruited engineer Andrei Karpathy to lead Tesla's Autopilot team. Meanwhile, Altman created a commercial division for OpenAI, received $13 billion in investment from Microsoft, and once again poached Karpathy.

OpenAI's ChatGPT4 was released in March 2023. Google also released a chatbot called "Bard" in response. The race to develop a product that can converse normally with humans and perform text-based intellectual tasks began in the form of "OpenAI/Microsoft" vs "DeepMind/Google".

It has developed into a serious situation that Musk was concerned about. With Microsoft and Google involved, he was worried that chatbots and AI would have political connotations. He was also concerned about being infected with the "WalkMind virus".

Furthermore, Musk believes that self-learning AI may one day rebel against humanity. It was conceivable that people would use AI chatbots to flood X (formerly Twitter) with hoaxes, prejudices, and scams.

That's why Musk needed to develop an AI that would solve such concerns. He believed there should be a third gladiator in the OpenAI vs. Google duel, and despite giving OpenAI $100 million in funding, he was eventually chased away.

In February 2023, Musk contacted Altman to bring OpenAI's founding documents to Twitter. He pointed out that it was illegal to turn a non-profit organization founded with donations into a for-profit organization that would make huge profits. Altman tried to convince Musk that it was legal, but failed. He proposed to create a new AI company and offer shares to Musk, but was rejected.

Musk sharply criticized, "OpenAI was founded as an open source (hence the "Open") non-profit organization to compete with Google, but it has become a closed source profit-maximizing organization essentially controlled by Microsoft. I don't know how a non-profit organization that donated $100 million can become a for-profit organization with a market capitalization of $30 billion."

Musk will never forgive Altman. That's why he founded xAI.

On the other hand, Altman is a sensitive person who doesn't like conflict. He has not made huge profits from OpenAI compared to Musk. However, while I understand to some extent Musk's idea of ​​AI rebelling against humanity, I don't think that's going to happen anytime soon. I think that AI will currently contribute more to humanity.

Musk believes that we should create AI that protects humanity and compete with OpenAI and Google. Above all, it is a great regret for Musk that he cannot join this AI war.

AI fuels data. GPT4 is trained daily with huge amounts of information. Both Google and Microsoft have huge amounts of data from search engines, cloud services, free email, etc. The amount of data that can be used will continue to increase enormously.

Mr. Musk uses X's posts as data. A total of 1 trillion posts, with 500 million posts being added per day. Posts of conversations, news, interests, trends, discussions, and videos are increasing in real time. This is the best live data available. This is the source of AI development.

In fact, there is a theory that the acquisition of Twitter is not because of Mr. Musk's love for Twitter, but because it can be used for AI. Of course, he is also thinking about monetizing Twitter after X. However, he thought that there were other ways to use it as a data source for AI, and that this would lead to the monetization of X.

Mr. Musk has another source of data. That is the 160 billion frames of video sent per day from Tesla's in-car cameras. This data is different from the text-based documents applied to chatbots, but is the scenery seen by humans moving around the real world.

This could potentially be used to create not just chatbots that generate text, but also AI for physical robots. Combined with X's data, it could create the most powerful humanoid AI robot. That could bring huge profits to Musk.

"How can we make AI safe?" Musk asks himself. "I've been thinking about this for a long time. What can we do to make AI less dangerous and keep human consciousness alive?" he said.

The new AI will perform machine learning on its own and take in information. It can teach itself what and how to output. It can also improve its own code and improve its capabilities. Mathematician John von Neumann and science fiction writer Verner Vinge named the moment when AI trains itself to move forward and leaves humanity behind as the "singularity." "That moment may come sooner than expected," Musk is concerned.

xAI's chief engineer is Igor Babushkin, a former AI employee recruited from Google's DeepMind division. He thought it had to be a big startup like Neural Link.

Musk's plan has just begun with the announcement of Grok. Musk hopes that there will be a world in which AI can answer the big questions for humanity, even questions about existence.

"It's a maximum truth-seeking AI. It's trying to understand the essence of the universe, and such an AI would want to preserve humanity, which is an interesting part of the universe." What does the future look like in Musk's eyes now?

Image: Iolite, Shutterstock

Profile

◉Akira Satomi
Former weekly magazine reporter. In charge of scandals and economic articles. Currently a freelance journalist, he writes for monthly and weekly magazines and web media. While he writes about crypto assets and other Web 3.0-related topics, he also writes about B-grade gourmet food such as ramen and izakayas.

Related articles

X to become fully paid "$2 per month" Elon Musk's ideas and the true purpose of the "X Evolution Plan"

Elon Musk to develop "TruthGPT" to compete with ChatGPT