Dec-08-22

Confronting Anxiety from Artificial Intelligence: Personal Experience

Open-source of TensorFlow

I am often asked, 'As AI progresses, do you feel anxious? If so, how do you deal with this anxiety?' Here are my thoughts on this question.

In 2016, AlphaGo defeated top Go player Lee Sedol, and in 2017 it defeated world number one player Ke Jie. Since then, no human player has beaten AlphaGo alone, making it my most anxious moment regarding artificial intelligence.

In the same year, my partner Vincent and I attended a translation conference in China. Participants told us there were over 100 machine translation companies in China at that time. Since Google open-sourced its machine learning engine TensorFlow in 2015, companies have developed their machine translations based on it. This not only saves redundant development waste but also overcomes many difficult technical obstacles.

That year, I vividly felt the anxiety of many translators who were worried about their job sustainability. The atmosphere of 'Luddism' spread on the internet. So, I turned my attention to other industries to see if there were similar situations.

It was then I realized that since the rise of the internet in the 1990s and its popularization, people have accumulated vast data, providing the nutrients for the rise of artificial intelligence. Concrete results began to emerge after 2015."

...

Jobs most affected by artificial intelligence

In the resurgence of artificial intelligence, the professions most affected are those involving image recognition, such as radiologists, because currently, AI is particularly good at image recognition. AI expert Andrew Ng once shared on Facebook in 2017 that a radiologist wrote to him asking if he should switch careers, sparking many discussions. Today, image recognition technology is quite mature, and even consumers can unlock their phones by "face-swiping."

There are also jobs that are easily influenced by AI because they deal with most data, such as accounting. In the United States, for example, a start-up company have launched tax software that can be directly linked to the U.S. government, so the routine tax-filing business of accounting firms is affected.

In fact, this part of the business was originally a crowded market, and accounting firms did not earn much money from it. But at least it was a stable income, and it could also maintain relationships with their clients and thus have the opportunity to upsell them with. But after processing through AI software, this business was shattered. Accounting firms naturally do not like this situation, but for many small and medium-sized enterprises, it is a blessing to pay a few tens of dollars a month to handle routine tax matters.

In Taiwan, finance is a highly regulated industry, so the development of financial technology (FinTech) is slower, and accounting firms will not feel the obvious impact of technology in the short term. However, according to data analysis, in the long run, financial technology will make accounting firms' tax-filing business more and more difficult to maintain.

...

Jobs that are less affected by artificial intelligence

Compared to images and data processing, dealing with voice is much harder. I'm sure you've had unpleasant experiences using voice input: no matter how many times you say it, AI doesn't understand what you mean, or the voice-to-text is full of mistakes, because voice recognition involving language is very difficult to handle.

For example, if you ask Siri a question, in order to answer your question, it and the massive infrastructure behind it have to do the following:

  1. Convert the sound waves you emit into text (voice recognition)
  2. Determine what you want to express with these words (natural language recognition)
  3. Generate an appropriate response (natural language generation)
  4. Let Siri voice the answer to you (speech synthesis)

Each of these steps is as difficult as climbing to the sky, and has troubled computer scientists for decades. It was only recently that significant progress was made through deep learning. But even today, AI is struggling with generating natural conversations.

That's why I often tell translators that if you really look at the impact of AI on other industries, you will know that language is a mountain that is extremely difficult for AI to cross. Two weeks ago, I was talking to a few computer scientists on Clubhouse, and the topic of Alpha Go beating Lee Sedol caused a lot of uproar in the translation circle.

After listening, a scientist said that Alpha Go used a huge amount of computer power to beat Lee Sedol, which was equivalent to thousands of first-class chess players against Lee Sedol alone, so it was unnecessary to scare yourself with that matter. "Only when a computer with a similar human body size defeats humans, artificial intelligence should be considered successful," he said.

Therefore, for language-related jobs, especially translation, there is still a long way to go for AI to replace humans. And this is also why some people say that the future of translation is not machine translation, but human-machine collaboration.

...

So how do I face the anxiety brought by artificial intelligence?

My anxiety about artificial intelligence began to ease around 2018, and has now completely disappeared. The reason for this easing is not because I see that other industries are "worse off" than us, so I feel a little relieved, but because I put this matter in the context of the development of human civilization for hundreds of thousands of years. When I look at artificial intelligence from the perspective of human history, I feel that my personal ups and downs are insignificant.

Before we appear, homo sapiens had experienced countless generations of predecessors and successors, inventing countless tools, and even inventing language and agriculture, so that I (we) had the opportunity to see the world today. After looking at the history of human civilization, I feel that I am, as Eileen Chang said, "In the endless wilderness of time in thousands of years, there is no earlier, no later, just catching up." I benefit from the efforts and sacrifices of countless predecessors and have already arrived to enjoy the achievements of those who came before me.

I am fortunate to have caught up with the rise of artificial intelligence before and during its rise, and even perhaps to see the world "after" its rise. This is thanks to the contributions and accumulated wisdom of countless human beings, and I am just lucky enough to be born today to witness the huge changes before my eyes. Therefore, no matter what the final result of this change is, I feel that I am very "lucky".

After looking at artificial intelligence in the context of human civilization, I know that the impact of this issue is no longer simply a matter of the existence of an individual, a profession, or an industry, but a question of the fate of humanity as a whole. Recognizing this is the reason why I am no longer anxious. In fact, not only am I no longer anxious, but I am even more interested in thinking about more profound questions: What makes a human a human?

In the past, this question seemed to be something only philosophers would think about, but now it is very likely that at some point in their careers, everyone will have to think about this question in different ways. After Lee Sedol was defeated by Alpha Go, he spent some time thinking about these "exsistence" questions. I believe that thinking about these essential questions is the real way for humans to face artificial intelligence.

In order to better understand these issues, I recently read some materials about quantum physicists talking about consciousness (the quantum physics of consciousness), and I was curious why there are physicists studying consciousness. Is consciousness not the domain of psychologists and neuroscientists? As a result, I have a better understanding of the argument of some well-known physicists that "we are very likely to live in a simulation", and this understanding has been helpful for me to think about my own life.

I was born in the era of the first winter of artificial intelligence (1974-1980) and came into contact with computers during the second winter of artificial intelligence (1987-1993). Now artificial intelligence is in the biggest development stage in history, and no one can say for sure how long it will continue to develop, or whether it will encounter a third winter. But I am sure that no matter how artificial intelligence develops, it will always follow the same pace as humans, and will continue to stimulate us to think about the most essential and profound issues.

Finally, I want to use a verse from the Buddha scriptures written by Wu Zetian (武則天), the founding empress of Zhou dynasty in ancient China, to summarize the personal significance of the development of artificial intelligence: "This unsurpassed, profound and wondrous Dharma, Which is hard to encounter in countless kalpas; Now that I can see, hear and am able to uphold it, I vow to attain the Tathagata’s true teachings." (無上甚深微妙法,百千萬劫難遭遇。我今見聞得受持,願解如來真實義) If my thoughts can help you a little bit in facing artificial intelligence, I will feel very honored.

career

作者

Joanne Chou

Joanne is an English-Chinese translator, a UI/UX designer, and the co-founder of Termsoup.