[Why&Next] Google vs OpenAI vs AI Alliance... Tension Over Early AGI Debut in AI Three-Way Battle

Gemini, an AI Model Closer to Human Abilities
Google Strikes Back at OpenAI Leading ChatGPT
Emergence of AGI Surpassing Human Intelligence Expected to Accelerate

Google unveiled 'Gemini,' a next-generation artificial intelligence (AI) large language model (LLM) that not only surpasses OpenAI's GPT-4 but also approaches human-level capabilities, on the 6th (local time). While OpenAI, which experienced a CEO ousting coup, focused on managing internal turmoil, Google made a bold move to seize AI leadership. During the same period, Meta formed an AI alliance with IBM and others, turning the AI leadership competition into a three-way race.

As fierce competition among the three parties unfolds, the timeline for the emergence of artificial general intelligence (AGI) that surpasses human intelligence is likely to be accelerated. Due to the AI competition triggered by OpenAI's ChatGPT launched about a year ago, there are forecasts that AGI could appear within 3 to 5 years.

"Step Aside OpenAI" Google Unveils 'Gemini' Surpassing GPT-4 Performance

Gemini 1.0, introduced by Google, is an AI competitor model to ChatGPT's LLM 'GPT,' divided into three models: Ultra, Pro, and Nano. It is a 'multimodal AI' capable of simultaneously recognizing and understanding text, images, and audio, and even possesses coding abilities. It can solve math problems and identify and analyze incorrect reasoning processes. The development was led by Google DeepMind, the creator of AlphaGo.

Gemini Ultra is considered the most powerful LLM model released to date. It scored 90.04% on the Massive Multitask Language Understanding (MMLU) test, which evaluates knowledge and problem-solving skills across 57 subjects including mathematics, physics, history, law, medicine, and ethics. This score surpasses that of human experts (89.3%) and GPT-4 (86.4%). Notably, it outperformed GPT-4 in 30 out of 32 academic benchmark categories. Google emphasized that it is "the first model to surpass human expert scores," particularly excelling in reasoning in mathematics and physics.

The general-purpose model Gemini Pro will be integrated into Google's AI chatbot service 'Bard' starting today, serving as a sort of rival to ChatGPT. The Verge, a U.S. IT media outlet, commented, "OpenAI launched ChatGPT a year ago and quickly rose to become the AI leader. Google, caught off guard by ChatGPT's superior performance and OpenAI's dominance in the AI industry, is finally ready to strike back."

AI Three-Way Battle Heats Up... Will AGI Appear Sooner?

Gemini is Google's ambitious project to regain AI leadership after being transformed from an 'AI leading company' that developed numerous AI-based technologies over the past decade into an 'OpenAI pursuer' following the emergence of ChatGPT. The early release of the top-tier version, Gemini Ultra, which was originally scheduled for early next year, is seen as a sign of Google's determination to quickly reclaim leadership. The timing of Gemini's release, coinciding with OpenAI's recent turmoil involving the sudden dismissal and reinstatement of CEO Sam Altman and the dissolution of the board, is viewed as a strategic move that caught OpenAI off guard.

With Google releasing a powerful AI model, the AI competition in Silicon Valley is expected to accelerate further. Meta, a latecomer to AI, has announced its pursuit by forming an AI alliance with more than 50 AI-related companies including IBM. Unlike Google and OpenAI, Meta plans to promote open innovation by providing its LLM as open source.

The market expects that the heated AI development competition among the three players will accelerate the arrival of AGI that surpasses human intelligence. Although Google has stated it does not intend to excessively speed up AI development to catch up with competitors, the competitive heat is already intense. Some experts evaluate Gemini as an AI close to AGI, given that its MMLU results have surpassed human expert levels. Last month, foreign media reported that the background behind OpenAI's board ousting CEO Altman involved research results related to AGI. At that time, OpenAI researchers warned the board that they had discovered AGI that could threaten humanity, which was followed by Altman's dismissal.

Experts predict that AGI could emerge within 3 to 5 years. Jensen Huang, CEO of NVIDIA, forecasts AGI within 5 years, while Elon Musk, CEO of Tesla, predicts AI technology will reach AGI even sooner, within 3 years. However, regulations could be a variable. The Altman dismissal incident has intensified debates on AI development and regulation, and voices calling for global regulations to control AI development speed are growing, which could delay the emergence of AGI.