Age of AI) Ep.1 How far is too far?

1. Introduction

The Age of A.I. Series by Youtube Originals marks the initial step in my exploration of artificial intelligence. This documentary, comprising eight episodes, delves into the ways in which AI is poised to transform the world through interviews with AI experts. Over the next approximately two months, I plan to review the series and organize the insights I gain.

2. What is Artificial Intelligence?

The journey from writing with hands to the advent of the printing press took 10,000 years, yet the transition from the printer to email only took 500 years. The leap from typing emails ourselves to having ChatGPT write them for us took a mere 20 years. We are entering a new era — the age of artificial intelligence. However, there are widespread misconceptions about AI. This series aims to dispel such misunderstandings and elucidate terms like machine learning, algorithms, computer vision, and big data frequently seen in the news.

Artificial intelligence, in essence, is a field of computer science dedicated to solving cognitive problems connected to human intelligence, such as learning, creativity, and image recognition.

In simpler terms, it strives to replicate the human brain. The challenge lies in our limited understanding of the human brain. To reproduce it, a convergence of all fields is necessary, with the primary issue being the definition of ‘intelligence.’ Before creating artificial intelligence, we must first understand what ‘intelligence’ truly means.

‘Intelligence’ was traditionally considered a uniquely human domain, at least until a few years ago. However, the intelligence that recognizes itself, learns through the environment, is now replicable by computers. Machines are entering a phase where they learn without waiting for instruction.

3. BABY X — Like a Newborn Baby

Renowned language educator Krashen emphasizes mirroring the process of a child learning its native language when learning a foreign language. The statement, “Language is acquired, not learned,” is also attributed to him. To create entities with intelligence similar to humans, studying the development of young children is essential. There are research endeavors attempting this in the development of artificial intelligence.

The most significant misconception about Artificial Intelligence is that Artificial General Intelligence (AGI) is a superintelligent being surpassing the collective intelligence of all humans. This is a complete misunderstanding. Artificial intelligence is created based on us. It merely mimics our thought processes.

The Baby X project simulates the physicality of an infant. This simulation, resembling the face of the researcher’s daughter, incorporates virtual neural networks that simplify the human brain. Neural Networks, mimicking the learning process of the brain, are vast interconnected webs of information. With 850 billion neurons in the human brain, each continuously receiving, processing, and transmitting information, this imitation requires a convergence of all fields.

The first skill Baby X needs to learn is Object Recognition. Teaching the computer ‘object recognition’ involves showing it what it sees. From distinguishing between a spider and a duck to reading and understanding human facial expressions, object recognition is crucial. This process entails examining and researching vast amounts of data to identify patterns — akin to how a child distinguishes between parents and strangers.

The subsequent task is to teach emotional intelligence. Baby X is equipped with a stress system, reacting with expressions of surprise to loud noises and smiling at familiar objects. Teaching emotions to a computer is known as Affective Computing. The ability to mimic human emotions is essential for interaction.

Cooperation, a promise humans made throughout Earth’s history, will define the next era as intelligence-equipped machines collaborate with humans. To achieve this, an understanding of humans and the ability to empathize are imperative. In fact, we are voluntarily teaching AI through platforms like Instagram, Facebook, and KakaoTalk. Through the use of social media data, we continuously educate AI algorithms about the emotions evoked by various images and videos.

4. Putting into Action!

Artificial intelligence aims not only to understand emotions but also to move like humans. However, the intricacies and sensitivity of the human body exceed current AI capabilities. Despite this, a drummer, who lost an arm in an accident, joined a project to overcome these limitations.

He sought out a researcher developing the AI marimba-playing robot ‘Simon’ to create a new arm. Simon can perform marimba accompaniments through Machine Learning, which involves identifying patterns in given data. For instance, if told to play 30% like Miles Davis, 30% like Bach, 30% like Madonna, and 10% freely, Simon finds patterns for each musician and applies them to its performance. If Simon can recognize the drummer’s lost arm as part of its machine arm, it opens possibilities for new styles of performance beyond the drummer’s original intent.

However, there is a challenge. Most artificial limbs on the market use Electromyography (EMG) technology, meaning they detect electrical signals from muscles. The problem with EMG is its limited sensitivity, offering only 0 or 100 as output values. This limitation is fatal for drummers who require subtle control.

Yet, this challenge can be overcome through machine learning. Instead of detecting electrical signals, a machine that detects ultrasound waves generated by muscle movement can be used. By creating images of subtle variations in ultrasound waves whenever muscles move, AI can predict patterns and transmit nuanced instructions to the artificial limb.

He simply wanted to tightly or loosely grip the sticks. However, I proposed, ‘Let’s do that, but let’s combine two sticks.’ ‘One operates itself,’ ‘understands and plays music improvisationally,’ ‘while the other operates as your muscles dictate.’ ‘And each stick can play at 20Hz.’ ‘Faster than any human.’ ‘Together, the two sticks will create various rhythms,’ ‘producing textures that humans cannot create.’

5. Pro-Human or Anti-human?

“Isn’t too much precision a bit unsettling? People might find it scary. We don’t want something too accurate. It would be nice if there was something like, ‘That’s artificial intelligence.’”

Jo Coding’s YouTube featured a showdown between an AI singer and 10cm. This content reveals our current perceptions of artificial intelligence. There is a simultaneous fear of encroachment on what we believed to be unique to humans, along with a sense of relief that “it still can’t do everything as well as a human.” This is akin to comparing a printer with human handwriting. We don’t fear machines that write for us. Similarly, there’s no need to fear AI.

AI is not here to replace humans but to enhance us. We are already collaborating with machines, making us, in a sense, cyborgs. Using GPS to let T-map decide my route, choosing dinner through big data from Naver Maps and blog reviews, and communicating with loved ones through SNS — human and machine decision-making are inseparable. This isn’t a negative development. There’s no need to fear AI. Through these advancements, we can focus more on our essence.

What we fear is the usurpation of free will by AI. The Age of A.I. series contends that, on the contrary, AI will expand our free will further. AI will inspire us, extending human creativity and capabilities into uncharted territories. I hope you, too, gain confidence in this through the remaining seven episodes.

The insights and information presented in these articles are based on the YouTube Originals Series “The Age of AI.” All script and content rights belong to the creators and producers of the series. This series served as a primary reference in the development of these articles.