Artificial intelligence everywhere

for ai to enter our homes, we must have confidence it will safeguard privacy.

Kim Byoung-pil

The author is a professor at the School of Business and Technology Management at the Korea Advanced Institute of Science and Technology (KAIST).

I suddenly wondered how many intelligent machines I had in my house. There is an artificial intelligence (AI)-enabled voice speaker that once used to be a fad. After early amusement, it is now hardly used. There is a robot vacuum cleaner that I recently purchased. I have to question its intelligence as it often stops after hitting a hurdle during a cleaning task.

There are several more home appliances that also list AI functions. But they have not been tested as I only use the primary functions of the appliances. Many have raved that AI could entirely change our lifestyle. But so far the ripple has been small.

So, where is this “real” AI? The supercomputers are mostly at work in data centers of large IT companies. They are powered by multiple arrays of processor cores and torus networks. To realize the awesome AI featured in publications, several expensive machines that cannot be afforded by common companies or households are required.

We nevertheless can connect to the internet to experience basic near-human software intelligence. My request is processed through the AI at a data center and answered through the internet. AI functions can be easily accessed when my computer and smartphone are connected to the internet. It is as easy as that. The function would be upgraded in line with the improvements of the AI capabilities in the data center, without any extra toil.

There is a downside to this convenience. The user must provide private information to the AI operator. Privacy can be impaired if the company loses or leaks the private information. It is why many institutions ban the use of ChatGPT and other high-level AI services.

It’s equally risky when used at home. To enable AI to bring comfort to our everyday lives, we must agree to it watching over our lives — when we return home, what we do there, and who visits our home. Many would not wish to surrender all this information about our private lives when there is no knowing where and to whom the AI takes this information.

Several years ago, the indoor security camera at an apartment was hacked and leaked. It is frightening to think someone could watch what takes place in my home. The fact that the camera has an AI function could cause anxiety.

For AI to become commonplace, there must be faith that it would steadfastly protect our privacy. Privacy protection should not be selective but essential. A function that keeps user information internally at home or at a specific company without delivering it to the internet has been gaining a spotlight. Many companies are opting to install a neural network at their computer data centers, or building an on-premise system.

The method can be applied at home. The AI function used at home and in home appliances can be trained to be kept there without storing sensitive information.

But this function can be costly. The machines to process high-level AI are very expensive and demand a lot of electric power. Few companies or homes could afford an internal AI infrastructure. A technology to miniaturize the AI function that can operate in a less expensive machine needing less power is needed.

The evolution is ongoing. Voice assistants of smartphones make a good example. Previously, the assistant had to send signals to the internet to process the voice demand. But now the AI chip embedded in the smartphone can directly access the voice demand. Studies show that an AI of much lesser scale can perform the essential tasks needed at companies and households.

The race in AI development has taken a new direction as the result. Developers so far competed to make a bigger and more powerful AI. But the future competition would be turning out AI that is competitive and reliable in performance at a lesser cost and power usage.

Only when such accessibility is achieved can AI become an everyday part of our lives. We must feel safe about the robots in our lives doing household chores and commanding them to do tasks on our behalf. Such changes can be possible only when there is confidence that AI would safeguard the information of its masters.

Translation by the Korea JoongAng Daily staff.