메뉴 메뉴
닫기
검색
 

SCIENCE

제 24 호 KaKao's Kanana: Innovation or Just Another AI Letdown

  • 작성일 2024-11-24
  • 좋아요 Like 0
  • 조회수 109
이수민

Kicker: SCIENCE

KaKao's Kanana: Innovation or Just Another AI Letdown

by Sumin LEE, Reporter

minnida_@naver.com

In the 21st century, when fierce competition is taking place in companies from various countries, technological advances are developing at a fast pace so that we cannot catch up. Among such developing technologies, AI technology, which is the most notable, is applied to more industries and can be easily seen growing according to the environment. Apple's "Siri," which is most famous for some examples, recently released a version with chatGPT through an ios beta update, understands users' languages better than now, performs tasks to and from specific apps and asks chatGPT to work on users' requests. In addition, Amazon's "Alexa" is a technology that provides information on music performance, news, sport and weather through a voice recognition system and helps control electronic devices with voice in conjunction with smart home appliances. It can also be used in apps and websites such as smartphones and tablets but it can interact through the app and has the function of learning and storing user preferences, interests, and tendencies. At a time when such various companies are actively utilizing the advantages of AI, Kakao, a Korean company,also introduce new functions using AI. In this article, let's take a look at Kanana, a new Kakao technology.



What is ‘Kanana’?

       

The announcement of Kanana and the character

https://www.donga.com/news/Economy/article/all/20241022/130268340/1



  

  ‘Kanana’ is a messenger that includes "AI" technology that acts as a friend and secretary in one-on-one chat rooms and group chat rooms on an AI interactive platform. It provides an optimization service based on all conversations and cumulative records that take place in the app. It is a platform that allows you to summarize documents, create test questions, recommend places and contents, announce schedules and preparations, and recognize and talk in real time without user requests. According to Kakao’s announcement, understanding each individual context and situation and presenting new communications using AI is expected to contribute to the expansion of loyal users of the Kakao service ecosystem through the structure that ‘Kanana’ takes on. Currently, Kanana's AI interactive platform "Kanana" is undergoing in-house closed beta testing (CBT) within the year and will enter CBT for general users in the first quarter of next year.


The Technique of ‘Kanana’

  There are many types of communication platforms, and although there are some usability overlaps due to the subdivision of communication, we expect that It is confirmed that it will contribute to the expansion of loyal users in the Kakao service ecosystem. The platform used an AI orchestration strategy. AI Orchestra is a strategy to use AI models according to service functions, such as orchestras that use multiple instruments. Using AI models with different sizes and learning materials can improve the speed of work and if collaboration with Open AI, a company with the GPT series, is actively carried out and activated together, it can save money by selecting multiple AI models like the Korea’s three telecommunication companies that actively introduced overseas AI technology. AI language model of ‘Kanana’ uses a function call called function call, which is essential for implementing services based on the language model by connecting external tools such as behavior instructions and APIs that cannot be performed on its own. Datasets that use this function are configured to evaluate function names and factor extraction accuracy, function call result delivery accuracy, and association detection with callable functions. While third-party datasets are built only around accurate function call message generation in language models.


The Problem of AI Service is Rising

  However, there are limitations to this technology. The most notable part is concerns about "Cannibalization," referring to the phenomenon in which a company's new products erode the market for existing flagshipproducts, which means overlapping usability with existing KakaoTalk, and Kakao CEO Jung Shin-ah says that Kakao Talk and interactive AI service ‘Kanana’ are very different, and Kanana's purpose is to expand new services. If personal information is developed using technology at the risk of leakage of personal information due to AI technology, the damage to personal information exposure due to hacking can increase. A case in point is the leakage of personal information from Kakao Bank, the same affiliate, in 2024. It is also called a "money-eating hippopotamus" because it consumes a lot of money due to reckless competitive development related to AI and energy problems such as carbon emissions are noticeable. In addition, ethical and legal problems with copyright of creations made with generated AI may arise, and it seems necessary to compromise these problems and understand and utilize technology.



Have you ever heard of the term "media literacy"? It is simply a combination of human communication tools and technology, media, and literacy, which is the ability to read and write, that makes it easier for us to get new information, which can affect our perspective on technology.In the 21st century, when individual individuality is more important than common trends, it is necessary for consumers to critically view companies that plan and conduct marketing with "differentiation" as their strength and compete with corporate branding. In addition, AI is changing our world faster with the development of technology. Therefore,We can lose our jobs and sometimes we are in a situation where we are not properly compensated for the damage caused by technology due to unspecified laws.In 2024, a new term, Homo Prompt, appeared. It refers to people with high AI literacy who can freely use AI services. Taking advantage of AI's characteristics, where the degree of completion of answers varies greatly depending on the level of prompt, a space where humans enter commands into a computer, it is sometimes evaluated that an individual's competitiveness will depend on their ability to use of AI from 2024 onwards. Givem the situation around us, shouldn't we accept it as it is and have an attitude that allows users and consumers to develop through utilization rather than being overwhelmed by it?


Sources:

https://n.news.naver.com/article/021/0002670883?sid=103

https://n.news.naver.com/article/029/0002914447?sid=105

https://n.news.naver.com/article/031/0000882981?sid=105

https://www.kakaocorp.com/page/detail/11253

https://n.news.naver.com/article/015/0005055290?sid=105

https://n.news.naver.com/article/003/0012889398?sid=105