[CES 2020] LG Electronics envisions AI that can reason

LAS VEGAS – It’s a small butler, hard as a tin but intelligent as a human.

“Good morning, did you sleep well?” says a robot as it suggests breakfast menus, recommending a light green salad to help reduce stress before your big presentation at work.

The AI robot knows your schedule, can even assume the state of your feeling and offer due action plan.

This is what a Level 3 AI will be able to do for humans, according to the AI framework envisioned by LG Electronics and Element AI, a deep learning startup based in Canada. The vision was unveiled during a press conference hosted by LG at the Consumer Electronics Show 2020 in Las Vegas on Monday. 

Speaking at the Mandalay Bay Convention Center in Las Vegas, LG Electronics’ Chief Technology Officer Park Il-pyung explained, amid a wave of AI-related ideas and concepts, it is important to share a structured framework for the development of AI across industries so that the company may create a meaningful impact on the lives of their consumers.

Joined by Element AI CEO Jean-Francois Gagne on the stage, Park discussed the differences in AI experience from Level 1 through 4. The four levels run from efficiency to personalization, then to reasoning and ultimately, exploration. These are anchored in significant jumps in technical capabilities and its applications.

Efficiency, the lowest level, is where specific device and system functions can be automated through simple commands, which is currently possible with most voice recognition and AI-enabled products on the market today.

An example of first-level AI is the LG ThinQ air conditioner that has a smart sensor to detect the presence of people in a room and adjusting the temperature and airflow accordingly, Park explained.

The next level, personalization, focuses on pattern learning to optimize and personalize device functions, Gagne said.

LG’s next-generation R9 robot vacuum cleaner will be able to learn from mistakes, such as getting stuck in gaps and corners by remembering spatial patterns, meeting the Level 2 capabilities, Park said.

In the near future, however, LG’s goal is to reach the third level of reasoning. 

The third level envisions an AI that uses causality learning through the collective intelligence of a system made up of different devices and services. 




By perceiving the cause of certain patterns and behaviors, AI at this level can better predict and promote positive outcomes for users, Park said.

“Here, we can leverage the diversity of our product portfolio because we’ll need many intelligent touchpoints that interact with the user and gather information to understand the ‘why’ and establish causality.”

Gange said his company has recently embarked on research into that level and it would that some years to be realized.

Exploration, the highest level, is “the ultimate destination for LG’s AI,” Park said. 

At this level, AI-enabled systems will be able to develop new capabilities through forming and testing hypotheses to uncover new inferences based on scientific method called “Experimental Learning,” enabling them to learn and improve, adding more value to users’ lives.

By introducing its AI platform LG ThinQ two years ago, the company has been making efforts to increase its presence in AI.

The company has used a third of its 2,000 square meters of exhibition space for a ThinQ Zone at CES 2020 to showcase home AI services, such as dining solutions with robots and options for virtually trying on clothes.

LG will also unveil its connected car platform at the show for the first time.

By Song Su-hyun (song@heraldcorp.com)
Korea Herald correspondent