About 220 employees have participated in the in-house AI education program to date; the initiative serves as a launchpad for building an AI transformation ecosystem as real-world workplace applications multiply
At 8:40 a.m. Wednesday, on the second below-ground floor of SK Telecom's SKT Tower, employees' hands were already moving the moment a task was handed out: find answers related to the commercialization of a fictional deep-sea species called "Cheongram Signal Whale" — a scenario designed to simulate a new SK Telecom business push.
No one exchanged a word with the colleague sitting next to them. The tables held no notebooks or pens — only laptops. Employees pulled up large language models including ChatGPT, Gemini and Claude on their screens, then attached URLs and reference materials to their prompts.
"I uploaded three URLs to ChatGPT, copied in all the responses, and then told it: 'Everything you need to solve is in the URLs, and the answers are below — find them and fill in the responses,'" one participant said, describing how they structured the prompt and expected output.
A single sentence fed into an LLM dramatically reduced the human burden of sifting through an overflow of materials, juggling multiple variables and handling unexpected complications.
By 8:56 a.m., it had taken less than 20 minutes from opening a laptop and launching an AI tool to finding the answer.
The scene unfolded Wednesday at the EBB AX Club — SK Telecom's in-house AI capability program — held at SKT Tower in Jung-gu, Seoul, offering a vivid snapshot of how AI has transformed the company's office environment.
The EBB AX Club is a program SK Telecom runs to help employees build practical AI skills. Participants use AI tools to tackle tasks modeled on real workplace situations.
The response has been strong. Since its launch in March, the program has run eight sessions through May, drawing an average of 30 participants per session and about 220 in total. Wednesday's session alone drew 84 participants — 34 in person and 50 online.
With AI transformation, or AX, ecosystem-building at the top of the corporate agenda, the EBB AX Club has served as the starting point for getting employees to actively use AI on the job.
Much of the participants' effort went into identifying the right LLM for each type of task — because different models produce different answers.
Many employees supplement the ChatGPT Enterprise accounts SK Telecom provides with personal subscriptions to Gemini, Claude and other tools.
In one example from Wednesday's session, Gemini failed to produce a useful answer because the URLs in the task were sample pages SK Telecom's development team had created specifically for the exercise. ChatGPT, by contrast, ran its own web search and arrived at the correct response.
Kim Su-ji, a manager on the Growth Team within the management strategy division and a non-developer participant, said she fed all the necessary materials into the models along with the URLs and asked them to work through the problem. "I ran three LLMs at the same time for a single attempt," she said. "ChatGPT was the most useful for today's task — Claude didn't really come through."
What employees learn in the program feeds directly into their daily work. Presentation decks, meeting transcription and real-time summarization, earnings tracking and repetitive administrative tasks are all now handled through AI.
Choi Mok-won, a manager on T Service Planning Team 2, said his team falls under an AI and digital transformation planning unit, so everyone is already on board. "One team member builds all presentation materials entirely in Claude, and meeting recordings go straight into AI for real-time summaries," he said. "The regional marketing division is going through dealer earnings management one by one and applying AX to each."
Kim added that the management strategy division alone has more than 50 people, and the strongest performers cross-check results across not just multiple LLMs but various other tools as well. "People end up focusing more on things that require human judgment — like looking at something from a different angle, or catching when the AI got something wrong," she said. "There's a real shift toward handing repetitive work off to AI, and the momentum has gotten much stronger this year."
ko@heraldcorp.com
