SK Telecom ends metaverse offering; Meta launches AI model

Getting your Trinity Audio player ready...

South Korean telecommunications giant SK Telecom (NASDAQ: SKM) has announced plans to shutter its metaverse platform ifland to increase the size of its bets on artificial intelligence (AI).

According to local news outlet Business Korea, SK Telecom will bring down the curtain on ifland in March 2025, ending five years of metaverse operations. Described as the end of an era for SK Telecom, the move signals an intention from the Korean firm to turn its attention to AI innovation.

Launched in 2020, ifland allowed users to explore virtual worlds with customizable avatars. At the height of its powers, the metaverse platform attracted over 4 million monthly active users, but waning interest in Web3 saw the figure dwindle to around the three million mark.

While the figures for ifland hit a slump, SK Telecom’s AI ambitions took off with a bang, with a range of early experiments netting large wins for the firm. SK Telecoms began pitching its tents with AI as early as 2022, riding the wave of ChatGPT and other generative AI offerings to roll out its large language model (LLM) in 2024.

“As a global AI company, we aim to concentrate our capabilities on AI,” said the company. One year after its foray into AI, the telecommunication company’s revenues have never been higher, netting $13.2 billion. New subscribers poured in from the South Korean and regional markets attracted by the “strong AI infrastructure and technology capabilities” it amassed over 24 months.

“A”—its AI agent—has earned critical acclaim from end users and can go toe-to-toe with global AI offerings in terms of capabilities. The company’s decision to shutter ifland will provide additional resources to power its AI arms race as it aims to keep pace with the rest of the pack.

Experts say that “A” will receive a raft of upgrades in 2025, including call summaries and multi-agent functionalities. For the first time, the AI assistant will be tailor-made for enterprises, and a new version is expected to be released for European and North American markets dubbed Aster.

Looking at the landscape in South Korea reveals a similar trend among technology companies. Web3 companies are shuttering their solutions in favor of other innovations in droves, casting a shadow of doubt on the industry’s future. 

KT has shut down its Genieverse and Meta Lounge, while Kakao has terminated its Colorverse platform. While Zepeto by Naver continues to run, it is unclear whether the company will mirror the actions of its peers to shift its weight to AI solutions.

Meta rolls out AI model to improve metaverse experience

In other news, Meta (NASDAQ: META), the parent company of Instagram and Facebook, has launched a new AI model designed to improve the movement mechanics of avatars in metaverse environments.

Dubbed Meta Motivo, the AI model is expected to assist users of metaverse platforms in achieving human-like movement, enhancing users’ overall experience of virtual worlds. According to an official statement, the offering is the “first-of-its-kind model” with broad functionality for developers.

With its open-source mantra, Meta Motivo will be available for developers to use at zero cost, a strategy designed to foster collaboration while gaining an edge against its peers. 

Use cases for Meta Motivo are endless, supporting the development of realistic movements in virtual worlds. The behavioral foundation model can be deployed in designing non-playable characters in gaming, but Meta says the model offers far-reaching use cases.

“We believe this research could pave the way for fully embodied agents in the Metaverse, leading to more lifelike NPC, democratization of character animation, and new types of immersive experiences,” said Meta in a statement.

An interactive demo on the website reveals that the model supports the design of whole-body tasks with zero shot interference without the need for finetuning. Users simply have to input their prompts and select an array of movement mechanics from a pool of options.

The model adjusts to changing environmental conditions, including wind and gravity, while offering unsupervised reinforcement learning. Per the announcement, the open-source demo comes retrofitted with a pre-trained model, a new humanoid benchmark, and a training code to simplify processes for developers.

“At test time, our model can be prompted to solve unseen tasks such as motion tracking, pose reaching, and reward optimization without any additional learning or fine-tuning,” read a report.

Meta is adopting a bifurcated strategy involving the emerging technologies of metaverse and AI, a play that has seen its capital spending surge to nearly $40 billion. The company has shown little to no signs of tapering its commitment to AI and the metaverse, rolling out new offerings for consumers on its social media platforms.

Going forward, Meta is set to roll out a new training model for language modeling in a valiant attempt to decouple reasoning and language representation. Dubbed the Large Concept Model (LCM), Meta says the model veers off a tangent from the traditional large language model and can identify “the next concept or high-level idea.”

In order for artificial intelligence (AI) to work right within the law and thrive in the face of growing challenges, it needs to integrate an enterprise blockchain system that ensures data input quality and ownership—allowing it to keep data safe while also guaranteeing the immutability of data. Check out CoinGeek’s coverage on this emerging tech to learn more why Enterprise blockchain will be the backbone of AI. 

Watch: Micropayments are what are going to allow people to trust AI