Korea’s Naver joins generative AI race with HyperCLOVA X large language model

One other large gun is coming into the AI race.

Korean web big Naver right this moment introduced the launch of HyperCLOVA X, its next-generation giant language mannequin (LLM) that delivers conversational AI experiences by a question-answering chatbot referred to as CLOVA X.

The corporate stated it has opened beta testing for CLOVA X in English and Korean and can make HyperCLOVA X accessible to enterprise customers, permitting them to customise the mannequin on their very own knowledge. It additionally plans so as to add an AI perform referred to as Cue into its search engine, very like what Microsoft has carried out with Bing, by November 2023.

The transfer comes at a time when firms throughout sectors are racing to construct AI into their inside workflows to drive efficiencies whereas the distributors offering these companies are going all-in on new capabilities to make the implementation simpler. Only in the near past, OpenAI, which began the generative AI wave, introduced that its GPT-3.5 Turbo mannequin can now be fine-tuned on enterprise datasets, whereas Midjourney expanded its device with a brand new generative infill characteristic.

What to anticipate from HyperCLOVA X?

HyperCLOVA X builds on its predecessor (HyperCLOVA), which has greater than 204 billion parameters. Naver hasn’t shared the precise variety of parameters for the brand new mannequin however it does word that it has discovered 6,500 instances extra Korean knowledge than OpenAI’s ChatGPT (powered by GPT-3). This makes the mannequin, and CLOVA X, notably helpful for localized experiences the place it might probably perceive not solely pure Korean-language expressions but in addition legal guidelines, establishments and cultural context related to Korean society to offer solutions.

“HyperCLOVA X … improves on the earlier LLMs, that are liable to giving hallucinations and fallacious info, and have shortcomings in offering up-to-date info or performing calculations,” the corporate writes on its web site. 

Naver additionally plans to make the expertise multimodal, in order that HyperCLOVA X will be capable of generate not solely textual content outputs however photographs, movies and sounds. For instance, customers might edit pictures “simply by attaching a file and chatting,” the corporate defined in a weblog publish, whereas noting that the performance can be added at a later stage.

Integration with enterprise techniques

Along with common use by way of CLOVA X, Naver’s new generative mannequin shall be open to customization by international enterprises. This can rework the generalist mannequin right into a specialist one, permitting groups to make use of it of their desired workflows, very like the best way OpenAI offers its GPT household of fashions. 

“You may tune HyperCLOVA X within the path you need by utilizing the information required by every trade group. Relying on the sphere you’re employed in, there are infinite potentialities reminiscent of ‘HyperCLOVA X customer support,’ ‘HyperCLOVA X coding,’ ‘HyperCLOVA X residence home equipment,’ and many others,” the corporate writes.

Within the space of ​​customer support, Naver explains, the mannequin might robotically classify and analyze buyer inquiries to assist brokers plan situations for coping with prospects. In advertising, will probably be in a position to create advertising phrases tailor-made to the traits of the corporate or present a abstract of promoting experiences.

The race is on

With the launch of HyperCLOVA X, Naver is shifting to tackle different main gamers within the generative AI race. These embody ultra-scale suppliers reminiscent of Google and Microsoft-backed OpenAI in addition to area of interest distributors like Midjourney.

The corporate has a 500-strong AI group and is working with Samsung Electronics to construct an optimized AI semiconductor answer, which is to be one-tenth the scale of its current one and provide greater than 4 instances the effectivity.

Based on McKinsey, generative AI might add $2.6 to $4.4 trillion yearly to the worldwide economic system. That’s excess of many international locations’ present GDP.