A forum was held to discuss all sectors of the AI industry, including daily life, digital advertising, and autonomous driving.

Machine learning, which enables artificial intelligence, has already become deeply integrated into our daily lives. In particular, it has been used in internet advertising for a long time.

"The information on the internet, once called a sea of information, has turned entirely into data," said Jung Soo-dong, CTO of Wider Planet, a company that manages online data and provides advertising. On Tuesday, the 26th, discussions on AI industry business took place at the Westin Chosun Hotel.

The 'AI digital advertising' revealed by CTO Jeong has proven its effectiveness compared to conventional advertising. Unlike machine learning, advertising learns people. The more data available, the higher the accuracy of the data. However, as interfering factors also increase, engineers end up using algorithms with lower accuracy to increase efficiency.

It recommends advertisements likely to generate good revenue based on click responses. Machine learning is being applied in almost every field, from ad requests to the exposure stage.

CTO Jeong stated that while 'feature engineering,' which involves selecting, refining, and curating training data, is particularly important, there are limitations to the areas where computers can perform this task, thus requiring more talent.
Regarding the autonomous driving technology levels announced by SAE, which have been controversial depending on the perspectives of those in the domestic industry, it was stated that ▲ Level 3 enables precise predictive diagnosis by realizing perception and judgment functions in the vehicle system, and ▲ Level 4 is a stage where accident response is possible through dual devices within the vehicle.

There are many questions about the technical difficulty of Level 3 and Level 4. Level 3 requires understanding numerous scenarios, which involves telling the driver to prepare for problems (if time permits). This is where artificial intelligence is needed. Level 4 is a problem for manufacturers to design the space for installing in-vehicle devices.

Citing the results of a 2016 survey conducted by the American Automobile Association (AAA), he stated that human trust in autonomous driving remains low. “While artificial intelligence is an indispensable element in moving toward the stage of fully autonomous driving, it carries uncertainty. There are still challenges remaining to resolve that 1% of uncertainty.”