[The Era of Supermassive AI] Jang Du-seong, KT Convergence Technology Institute Lead: "Innovation with 'Large AI' that Learns Like Humans"
'Large AI' Leading Innovation in KT's Core Business
Applied to Customer Center, Reducing Average Consultation Time by 15 Seconds with More Natural Conversations
Jang Doo-sung, head of KT AI2XL Research Lab, is giving a lecture on the theme "Large AI To Everywhere MI:DEUM" at the "Get Ready Web 3.0, The 1st Mega AI Ecosystem Opening" forum hosted by Asia Economy on the 27th at Gravity Seoul Pangyo in Seongnam, Gyeonggi Province.
Seongnam=Photo by Kim Hyun-min kimhyun81@
[Asia Economy Reporter Seungjin Lee] “The goal of teaching language like a human was the beginning of the next-generation artificial intelligence (AI) 'Large AI.'”
Jang Doo-sung, head of KT Convergence Technology Institute AI2XL (AI To Everything Lab), stated this on the 27th at the Asia Economy 'Get Ready Web 3.0, 1st Mega-scale AI Ecosystem Opening' forum held at the Gravity Hotel in Pangyo.
Existing AI had limitations due to its learning method relying on continuous construction of learning data and manual work by humans. To apply AI to new businesses, it was necessary to completely retrain the AI in the language related to that business, which incurred significant costs and time. Accordingly, KT AI2XL developed 'Large AI' applying self-supervised learning. Self-supervised learning is an AI technology that finds and analyzes rules by itself with minimal data, recognizing objects and assigning meaning without separate human guidance.
Regarding this, Jang explained, “The previous AI learning method was like underlining a book and telling what each part means,” adding, “'Large AI' with self-supervised learning can learn all the books in the world without manual work.”
'Large AI' is already leading innovation in KT’s core businesses. KT operates the largest customer center in Korea. 'Large AI' leaves memos of conversations between counselors and customers and stores previous discussion details to be used in follow-up consultations with the same customer.
Jang said, “Previously, sentences were extracted from stored consultation content to provide summaries,” and explained, “When using 'Large AI,' we are preparing to summarize the same content into natural sentences focusing on key points, expecting to reduce consultation time by more than 15 seconds.”
Hot Picks Today
"Rather Than Endure a 1.5 Million KRW Stipend, I'd Rather Earn 500 Million in the U.S." Top Talent from SNU and KAIST Are Leaving [Scientists Are Disappearing] ①
- Brilliant Korean Technology Flows Overseas... Subsidies Granted, but "No Product Launch Allowed"
- "If That's the Case, Why Not Just Buy Stocks?" ETFs in Name Only, Now 'Semiconductor-Heavy' and a Playground for Short-Term Traders
- Small Business Community Urges Samsung Electronics Union to Halt Strike, Citing Economic Damage
- "No Cure Available, Spread Accelerates... Already 105 Dead, American Infected"
KT is also applying 'Large AI' in various other fields. They plan to innovate by applying it to AI TV 'Genie TV,' AI speaker 'Giga Genie,' AI hotels, and intelligent robots.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.