Do You Know Me Better Than I Do? A Better Life Provided by Artificial Intelligence (AI)
Jul 30, 2020
Have you ever experienced artificial intelligence?
Does the image of ‘J.A.R.V.I.S.’ from Iron Man come to mind, or perhaps the legendary Go match where the computer program AlphaGo competed against world champion Lee Se-dol? Though we've undoubtedly witnessed many stories featuring AI's role in modern society, how would we describe our own experiences with AI in real life?
Encountering AI in Everyday Life
AI has been featured in unforgettable ways in films made famous all across the globe. Popular examples include the robots and computer systems of the Marvel and Star Wars franchises, not to mention Ex Machina, I, Robot, and the Terminator series.
We're introduced via screen to artificial voice assistants who can communicate in natural language, androids made with strikingly human features, robots with emotions just like humans, and even robots who want to be human.
Could it be then, that AI only exists in the realm of sci-fi?
We tend to view these cinematic flights of fancy as hallmarks of a distant future. However, as I mentioned back in Part 1, artificial intelligence is an old technology. AI affects our daily lives whether we know it or not, and we can find evidence of it in the systems, services, and devices we use every day.
Whether it's a camera that automatically recognizes vehicle numbers at the entrance of a parking lot, or the text messages we send using speech recognition from our smartphones, we are currently experiencing AI as our past selves could only imagine.
Development Levels of AI, As Proposed by LG Electronics (AIX)
How will artificial intelligence– as seen in movies– develop in the future?
As AI continues to make great strides, we might imagine the direction of its advancement to head in the direction of developing technology to solve difficult problems. However, LG Electronics is focused not only on the technology itself but on the human experience alongside technology, for the purpose of creating a better life for all. We wish to present a clear and structured road map for the whole industry, targeting the meaningful growth of artificial intelligence with systems like LG ThinQ.
Here, we outline the development levels of technology as changes in the average user's experience in autonomous driving. Although the actual differences between the levels are far more complex, the below example illustrates the levels at which the average user can free their eyes, feet, hands, and thoughts from driving a vehicle.
Automated driving systems (ADS) is a standard of development in vehicular automation, with different levels ranging from no automation needed, to full-operation automation. With the field of AI seeing new innovations emerge every year, further research and development could harness its growing capabilities to automate simple– and not so simple– tasks.
LG Electronics and Element AI have collaborated on defining development levels to share with the AI community. Each level has been carefully established from the perspective that the future of AI should be designed with a human-centric focus. We named this as Levels of AIX the AI Experience.
※Element AI: An AI company co-founded by Yoshua Bengio, a professor at the University of Montreal, and founder of Mila, an artificial intelligence institute in Quebec. Bengio is a world-renowned scholar in deep learning and artificial neural networks.
The levels of AI (AIX) were presented at the LG Press Conference at CES 2020 by President Il-Pyeong Park, CTO of LG Electronics, and Jean-François Gagné, CEO of Element AI.
The development structure Levels of AIX introduced by LG Electronics consists of 4 levels: Efficiency (Level 1), Personalization (Level 2), Reasoning (Level 3), and Exploration (Level 4). Let's take a closer look at each level.
Level 1 (Efficiency): Your Wish is My Command
In Level 1, AI can perform tasks efficiently, having drawn from large sets of data to recognize patterns and apply predictable solutions. The tasks can range from simple, routine tasks to far more complex ones difficult for even humans to do.
AI can be used to read medical data or X-ray images to inform doctors of suspicious areas and assist in prescribing medical treatment. It can view vehicle license plates at the entrance of an apartment parking lot to judge whether or not they are residents, and controlling access accordingly. It can employ speaker recognition to provide relevant information and even turn on the lights in your house by voice command. The scope of its assistance can be widened or narrowed to enable humans to work with AI in the way that is most valuable to them.
Because artificial intelligence is a machine, it can do repetitive tasks without getting bored, and because AI learns from a database of ever-growing information, it can recognize logical differences across numerous cases and process a veritable load of information in a matter of seconds.
Collaboration with AI allows people to focus on value-added work and reduce judgment errors caused by human fatigue or stress.
A washing machine can be made more efficient with AI by being able to determine the amount of clothes in a load, match it with the appropriate amount of water and detergent, wash in the direction that most minimizes damage depending on fabric type, and adjust washing settings depending on soil level.
Level 2 (Personalization): Shh, Don't Speak... I Already Know
The value of a "me"-centric experience can hardly be quantified. For example, a parent may very well prefer a pediatrician in their own neighborhood, one who has been treating their child since infancy, over a renowned doctor at a distant university hospital, who runs down a checklist of symptoms and prescribes treatment according to textbook medical knowledge. This could be because of the stark difference between that overly clinical setting and a familiar office where the parents are recognized and greeted upon arrival, asked about their recent health, and where the doctor provides a diagnosis based on both the child's current symptoms as well as their past experiences.
The second level of AI can be personalized so that pattern-learning is made possible through accumulated interactions with the user. AI-enabled appliances can analyze users' past behavior to identify patterns as a predictor of future behavior. Even if a product or service has multiple simultaneous users, speaker recognition allows AI to analyze each voice, face, and usage to extract meaning from unique patterns, recognize the user without their needing to log in, and operate accordingly.
When a user says, “Please recommend a recipe for spicy food," an AI refrigerator belonging to Level 1 can recommend a general recipe without considering the user's previous experiences. However, an AI refrigerator that belongs to Level 2 understands what food the user has enjoyed in the past and recommends recipes suitable for the situation. In addition, AI can remember a family's favorite beverages, order them at the right time before they run out, even buy them at a cheaper price when needed.
If your floor-cleaning robot learns the layout of your house and the amount of dust typically collected in a regular period, it can reduce cleaning time more and more with each use, and tidy up your house more efficiently.
Level 3 (Reasoning): Analyzing Cause and Effect
Humans analyze situations based on multiple data points.
A café owner can predict that a customer will likely order an iced beverage by observing the day's weather, the perspiration on the customer's forehead, or their hand fanning themselves as they enter. A doctor can diagnose a patient who has caught a cold and prescribe medication based on the symptoms of a runny nose, a slight fever, and a swollen throat. Humans can judge or predict the outcome of a situation based on acquired knowledge, even without statistical analysis of the specific situation at hand.
AI technology has not yet reached this level, so it judges based on specific knowledge that humans have accounted for in advance. The AI expert system was built upon this notion. However, the third level of AI development is less about humans creating this specific knowledge in advance and more about AI systems building knowledge independently and reasoning through self-learning from existing data sets.
The third level of AI analyzes the use of various products and services through cause-and-effect learning and determines specific patterns and causes of behavior. Based on this, AI can predict the user’s needs and operate with efficiency, even when encountered with a new situation. In order to provide customized services to users, it is essential to collect information through various points of contact. While the second level of AI figures out individual correlations such as user-product and user-service, the third level understands a comprehensive cause-and-effect relationship by integrating the information collected from different products and services.
For example, if a user turns on a water heater, takes out thick clothes from a sensor-equipped closet, and prepares hot coffee, AI in the “Reasoning” stage determines that the user's actions are for the purpose of increasing body temperature. When the weather report forecasts a significant drop in temperature, AI can stop running the air conditioner, prepare the house for heating, suggest that users wear thick clothes, and ask, "Would you like to drink a hot coffee?”
Level 4 (Exploration): The Smarty Pants Studyholic
The last level is Exploration.
Athletic coaches try various methods to increase their athletes' capabilities in order to achieve goals and win medals in games like the Olympics. Instead of following methods that others employ or sticking to well-known techniques, a coach might establish a hypothesis regarding potential methods suitable for the athlete, and consult with them to determine the optimal method through trial and error.
The fourth level of AI establishes and verifies a hypothesis by itself, like a coach, and proposes a better solution. This level enriches a user's life through the practice of experimental learning, where AI employs logical reasoning to establish and verify a hypothesis to find better solutions to frequent issues. As AI discovers new ideas and acquires knowledge continuously, it repeats meaningful experiments so that new information can add value to the user's life.
If an AI system recognizes that a user sleeps most comfortably when the ambient temperature is about 17°C/63°F, it can offer help by circulating cool air via the ceiling fan to maintain that temperature during the user's bedtime. AI may ask, “Should I turn on the fan when you go to bed?” In a smart city based on fourth-level AI, we can build safer, more efficient communities by collecting information from vehicles and traffic sensors to optimize the local transportation system.
For more information and examples of AI development levels, please refer to the LG ThinQ Platform site linked in the image below.
What does the future look like for you? With leading experts in the field, LG Electronics is committed to developing new innovations in AI technology to provide you with a better life, and a better future for generations to come.
*This post reconstructed the post contributed to LiVE LG, https://social.lge.co.kr/technology/live_with_ai_02/.