Anthropic, an AI startup, has clinched a substantial $100 million investment from SK Telecom, a prominent player in South Korea's telecommunications sector.
This alliance aims to forge a multilingual AI paradigm.
SKT's CEO Ryu Young-sang expressed:
"Through the fusion of Anthropic's robust AI prowess and our Korean language-based LLM, we anticipate a symbiotic synergy. Our global telecom partnerships stand to be further invigorated."
Anthropic's collaboration with SKT is geared towards creating a multilingual LLM (Large Language Model), tailored to cater to diverse linguistic nuances.
Dario Amodei, CEO and co-founder of Anthropic, affirms that sector-specific LLMs carry the potential to establish safer and more dependable AI technology applications.
What is a LLM?
An LLM stands for "Large Language Model," which is a type of artificial intelligence designed to understand and generate human-like language.
In simpler terms, it is a smart tool that learns from a vast amount of text data to communicate and respond in a way that makes sense to us, humans.
In the case of Anthropic, they are teaming up with SK Telecom to create a multilingual LLM.
This means they're working on an advanced AI system that can understand and interact in multiple languages like English, Korean, German, Japanese, Arabic, and Spanish.
Anthropic's Claude
Launched in 2021, Anthropic's notable contribution, Claude, an AI system, streamlines various corporate functions.
Claude's utility extends to customer service, marketing, sales, and interactive consumer interfaces within the telecom sector.
Its underpinning technology leans on a "constitutional" training method, diverging from RLHF (Reinforcement Learning from Human Feedback) techniques. This distinction reduces susceptibility to human biases.
Anthropic's recent update to Claude Instant bot, version 1.2, offers a cost-effective alternative to Claude 2.
The comparable prowess to GPT-3.5 positions Claude Instant as a transformative advancement.
Notably, Claude AI processes over 100K tokens of context, surpassing OpenAI's chatbots, which typically handle between 4K to 32K tokens.
Tokens, serving as fundamental units of information in LLMs, are distinct from words or letters.
For instance, "A" or "Cat" could be tokens, while "Medicine" comprises three tokens and "Humanoid" encompasses two.