[Q&A] Yoo Young-sang, SKT CEO: "We Will Partner with More Than Half of the Companies Excelling in AI"

Telco Alliance's 'A.Dot' Global Expansion
"Fairness Issues in Network Usage Fees"

[Barcelona=Asia Economy Reporter Oh Su-yeon] Yoo Young-sang, CEO of SK Telecom, is accelerating the transition to an 'AI Company' by building the 'K-AI Alliance' with companies possessing AI technology. The plan is to collaborate with anyone who excels in AI.

On the 26th (local time), Yoo Young-sang held a press conference in Barcelona, Spain, where Mobile World Congress (MWC) 2023 is taking place, and unveiled the vision of 'AI to Everywhere.' Using SKT's AI services and technology, he intends to lead innovation centered on five major areas: ▲Customer & Technology ▲Spatiotemporal ▲Industry (AIX) ▲Core BM (Core Business Model) ▲ESG (Environment, Social, Governance).

Since his appointment in 2021, CEO Yoo has been emphasizing the transition to an 'AI Company' and has been focusing on expanding AI services such as A. Dot. On this day, he announced the 'K-AI Alliance,' formed by joining forces with companies possessing AI technology, signaling leadership in the AI ecosystem. Yoo said, "Our goal is to gather more than half of the AI experts in Korea into the K-AI Alliance."

SKT CEO Yoo Young-sang is holding a press conference on the 26th (local time) in Barcelona, Spain.
[Photo by SKT]

Below is a Q&A with CEO Yoo.

- AI is becoming a trend like a syndrome, and amidst this, Yoo Young-sang, CEO of SKT, is taking the lead. What prompted and motivated the emphasis on transitioning to an AI Company? Also, A. Dot has been in beta service for nine months; how do you evaluate its performance since launch? Beyond the MNO service assistant, what kind of personalized services do you expect it to grow into?

▲ Yoo Young-sang: In the past, many asked why we are transitioning to an AI Company, but now they ask more about how rather than why. Still, there are various reasons why a telco (telecommunications company) would transition to an AI Company. From a service perspective, telcos have many customers but very weak customer touchpoints. Customers see big tech or Over-The-Top (OTT) services as their touchpoints, not the telco. Telcos are increasingly losing contact with customers and thus losing value. How to regain that touchpoint is not easy through conventional means. We have launched OTTs and various services, but we couldn't beat existing OTTs. The advantage of AI services in the AI era is that Microsoft (MS), with only 3% of search share, can challenge Google's over 80% share. Because AI services can skip stages, companies that directly provide AI services can immediately engage with customers. This is one reason.

Another reason is that our business itself has not fully undergone digital transformation (DT), but ultimately, applying AI increases productivity and customer value. Even TVs become AI TVs, refrigerators become AI refrigerators, and their prices rise. We plan to apply AI to all our existing services. Another aspect is business-to-business (B2B). B2B businesses include cloud and others, but the core is AI and AIX. Some call it digital transformation (DX), but I think AIX is a more advanced concept. If we believe AIX can bring new value to industries, tremendous capabilities are required. If we have these capabilities, a telco can redefine the game by transitioning to an AI Company based on telecommunications. We consider this from a corporate strategy perspective.

A. Dot started from this perspective, but since we are on a path others haven't taken, there have been many trial and error experiences. One representative example is that customers are familiar with chatting on smartphones but not with voice-based communication. However, for us, voice is more important. Currently, half of our customers use voice. Customers are changing. We have put a lot of thought into how to incorporate character personas from both technical and humanities perspectives. These aspects are gradually being organized. With the emergence of ChatGPT, we conduct emotional and goal-oriented conversations, but we are now considering how to enhance knowledge conversations to ChatGPT levels. If all these are resolved, it will become a completely new and unprecedented service. If we implement this in alliance with global mobile network operators (MNOs), it will be a new service. The problem with existing MNO services is that, like One Store and Wavve, even if they do well domestically, they cannot achieve global scale, which is a disadvantage. Therefore, survival is difficult. We aim to create global-scale services through a global telco alliance. In the AI service era, telcos will not be suppressed by existing big tech. From this perspective, although A. Dot is still lacking, we continue to invest.

- You mentioned A. Dot's global expansion, but considering A. Dot as an AI service, there might be limitations in global expansion alongside legacy telco companies. Could you explain the specific alliance model and challenges?

▲ Yoo: Our global strategy is not yet perfect. Metaverse Ifland has gone global, but A. Dot cannot do so easily. Ifland can be launched globally by uploading it to the cloud. Cultural and linguistic characteristics are not dominant there. A. Dot faces many cultural and linguistic challenges. We could try to do everything in each country, but rather, we will proceed through a telco alliance. For that, we need a Large Language Model (LLM) that everyone can use, called a telco-specific model, but the A. Dot model alone is insufficient. We need to quickly collaborate with telcos to build this. Each telco can then take it and launch it according to their language and circumstances.

AI technology alone without services at the service layer does not make a service. Each telco has various services, some of which we have created. For example, bringing a popular service in Korea to the United Arab Emirates for immediate use, or enabling a popular service in Germany to be used in Korea through a platform. If these two are created, I think there will be progress in creating a much more global scale than existing telco services.

- At this MWC, GSMA seems to have delivered a message about fair network investment sharing, and SK Broadband is also in litigation with Netflix. What is the telecom industry's response to global content providers' (CP) free-riding on networks?

▲ Yoo: Network usage fees are widely discussed at MWC, but it should be viewed from a fairness perspective. There have been debates on net neutrality and other issues, but now the question is what is fair. To redefine, it is about how much role-sharing should exist between CPs and Internet Service Providers (ISPs). Approaching this with fairness logic rather than power will allow us to find solutions.

- You mentioned combining Phantom AI's autonomous driving full-stack solution with Sapeon’s autonomous driving chipset. The next model X330 is coming out in the second half; when will the autonomous driving chipset be released, and will it be sold to B2B car manufacturers?

▲ Ryu Su-jeong, CEO of Sapeon: X330 will be released in the second half, and next year X340, an autonomous driving ADAS chip, will be launched. We have created a roadmap to deploy solutions accordingly and will generate synergy within that. As you mentioned, we are working extensively with Tier 1 and B2B, and we also plan to build business with Phantom AI.

- The AI Contact Center (AICC) market is growing rapidly. I understand that a cloud-based AICC solution is being launched in partnership with SKT. Could you provide details? Also, has ChatGPT been introduced into this?

▲ Lee Han-joo, CEO of Bespin Global: SKT has an AICC solution and is independently seizing business opportunities to accelerate the AICC business. Our HelpNow team is also involved. HelpNow designs chatbots and operates the platform. We are jointly pursuing business opportunities and specific projects. As you know, ChatGPT has emerged, and MS is preparing for commercial release. We prepared in advance and integrated ChatGPT into HelpNow. The first commercialized service on MS Azure will be launched, and since HelpNow and ChatGPT are integrated into the specific AICC business opportunities with SKT, they will be naturally released together upon commercialization.

One more thing: such business opportunities will continue to arise. Companies not using AI will be left behind, and telcos have many roles to play. Another major MWC theme will be sovereign cloud. As AI rises in each country, each nation needs to have its own AI engine, and who will take charge?whether it will be telcos?is a big topic. SKT leading this can serve as a reference, and Korea has an opportunity to take the lead in the global telco market through the sovereign cloud trend.

- Driving AI inevitably raises ethical issues. OpenAI, which created ChatGPT, faced bias issues at GPT-3 launch. Korea also had the Lee Luda incident. How does SKT respond to ethical issues?

▲ Yang Seung-hyun, SKT CTO: We already comply with various safety guidelines and apply safety functions. As we expand datasets, we will continue to develop technology to mitigate hallucination issues like ChatGPT.

▲ Yoo: Regarding safety, when we developed A. Dot, we applied very strict safety measures, which made it less fun. If something slightly odd appears, it responds that it cannot answer. Overapplying safety can be problematic. Setting the boundary is somewhat an art.

- Recently, the government diagnosed the telecom market as an oligopoly and announced plans to induce competition. From SKT's perspective, there may be parts to comply with government policy and parts that are regrettable.

▲ Yoo: Regarding government policies, we will do our best to follow government measures on communication fees as part of various livelihood policies. For example, we will try to quickly negotiate with the government on senior and mid-tier plans to reform tariffs.

- SKT emphasizes AIX, but the AI service mainly pushed is the B2C A. Dot. Since advancing A. Dot requires significant investment, revenue generation is necessary long-term. You mentioned customer touchpoints, but it seems it won't be just about data acquisition. What kind of services do you want to create long-term? In a previous conference call, you mentioned growing A. Dot as the next internet. Will it be a portal for advertising business, or if ChatGPT is integrated, could it be a tool for productive work? What business model do you plan long-term?

▲ Yoo: I don't think there is no business model, but we haven't imagined much yet. First, we need to secure customers. We are creating Jarvis and Samantha; Jarvis helps with various tasks, and Samantha acts as a friend when lonely or needing companionship. If these roles are well fulfilled, what willingness to pay might exist? If done well, whether directly or indirectly, through subscriptions, advertising, or commerce, I believe there will be sufficient willingness to pay. If each customer has one agent, and every citizen has one agent, the relationship with that agent will be stronger than any other. With such a strong relationship, I believe there will be significant willingness to pay. How many customers use it and how much they use it is more important. Then the business model will naturally emerge without much worry.

- No matter how good a game console is, it cannot succeed without fun games. I think the K-AI Alliance is like various games. Since most B2B companies are involved, how far do you plan to expand into B2C or K-content? Also, when talking about ChatGPT, I felt like you said, 'See, when I talked about AI, no one believed it, but the AI era has come.' Was it difficult to persuade employees? Were there challenges in changing the organization? Lastly, do you think the recent ChatGPT craze will be a one-time phenomenon or continue?

▲ Yoo: It may vary by person, but although ChatGPT is the most recent, I believe AI has potential comparable to other technologies like the internet and mobile. ChatGPT may fade, but AI will not. There will be no major variable against AI continuing.

Changing the organization into an AI Company is not easy. The first difficulty is persuading why it should be done, but with ChatGPT, it has become much easier. People no longer ask why but how. I advocate AI across the entire organization, using traditional change management tools with KPIs and other tools to AI-enable the entire organization, including AI-utilizing and AI-developing teams. I hope to achieve results this year.

K-AI Alliance is my ambition. I think Korea has many AI experts compared to other countries but is still far behind the US and China. My goal is to gather more than half of Korea's AI experts into the K-AI Alliance.

- ChatGPT was released last year, but the recent hype is largely due to users sharing experiences virally. The general public is not familiar with A. Dot or KT Digico. What is your strategy to approach the general public?

▲ Yoo: We must create it. I have experience succeeding with Cyworld, Melon, and T-map. Although I have more failures, I remember valuable services spreading by word of mouth. We will definitely create such a killer service to enable viral spread.

- Wherever AI is applied, the fastest benefit customers feel is likely in fees. Do you plan to use AI for personalized fee recommendations?

▲ Yoo: We are already doing that. The prerequisite is to diversify plans to broaden customer choice. Although there are difficulties, with the upcoming mid-tier plan reform, I think we can do more in this area.

- You mentioned Sapeon was valued at 500 billion KRW and received investment. From whom did you receive investment? If disclosure is difficult due to negotiations, could you at least say whether the private equity or venture capital is domestic or foreign? When Sapeon was valued at 500 billion KRW, did the value of existing SKT, SK Square, and SK Hynix shares change? How did it change?

▲ Yoo: Once finalized, we will disclose details separately.

- I have used A. Dot, but ultimately you have to open the app. I think this is a disadvantage compared to Siri, Bixby, or Hey Google. Do you plan to discuss with platform operators to resolve this? Currently, it is used within smartphones, but do you have plans to expand to other devices or services?

▲ Yoo: Google and Apple have operating systems (OS). Apple and Samsung have devices. Compared to them, telcos are absolutely disadvantaged in this regard. Because of this disadvantage, what we do as much as possible is pre-installation. Just being pre-installed is significant. Non-telco apps cannot even be pre-installed. When smartphones first came out, A. Dot was pre-installed. If there is a better way, that would be great, but this is the reality. Voice interfaces fit much better on devices other than smartphones. We are working on smartphones, but ultimately, I believe the mobile revolution will manifest as a device revolution. New devices have not yet appeared, but for example, in cars, voice interfaces like Nugu Auto are most effective. When (AR/VR) glasses come out, obviously, there will be no keyboard or touchpad, so voice will be used. Future devices will have much higher potential for conversational UI/UX. Therefore, although currently limited to smartphones, I think A. Dot's competitiveness will be better demonstrated in the future.

- A. Dot seems to be core to the AI vision. You said customer acquisition is important. What is the subscriber target within this year?

▲ Yoo: Since it is currently a beta service, the key question is when to launch the commercial service rather than subscriber targets. Once commercial service starts, subscriber targets can be set.

- Due to issues like ChatGPT, although business areas differ, SKT must compete with Google, MS, Naver, Kakao, etc., in AI services. Actual users need to try SKT services and feel AI. In this regard, what differentiates SKT’s technology, competitiveness, and vision from global tech companies?

▲ Yoo: First, why did SKT, among global companies, launch AI services first, even if not perfect? Regarding current services, we are not legacy, but they are. They have much to lose as service providers. We are a telco but have few services, so we could enter AI services first. It’s not that our AI technology is superior, but because we have nothing to lose, we could enter first. Regarding technology, I think telcos need to create LLMs they can do well. We need vertical LLMs tailored for telcos. If we go fully general language models, I doubt we can beat Google or MS.

- You said Korea’s AI is far behind, and you want to collaborate with half of domestic AI companies. Is there a possibility of cooperation with the second-largest telco? Netflix’s new CEO is coming to MWC, and Vice Chairman Park Jung-ho sometimes had coffee with Netflix’s former CEO. Do you plan to meet?

▲ Yoo: It’s not easy, but I think companies that are among the top ten AI companies domestically should cooperate in some form. We are willing to cooperate. I don’t know the new Netflix CEO well since the CEO changed. I used to have meals with the former CEO, but I don’t know how it will go.

This content was produced with the assistance of AI translation services.