[Opinion] The Fired Artificial Intelligence

Im Joo-hwan, Honorary President of the Korea Communications Society

Six years ago, Lee Sedol 9-dan played a Go match against AlphaGo, an artificial intelligence developed by DeepMind, a subsidiary of Google. Go is a game that makes the highest use of human advanced intuition and instinct. Defying expectations that Lee Sedol would win, AlphaGo achieved victory with 4 wins out of 5 matches. Amid tremendous shock, an AI craze swept across the Korean Peninsula.

AlphaGo was upgraded in 2017 to ‘AlphaGo Zero.’ The hardware remained the same as AlphaGo’s, but the software was changed. Within 36 hours of starting training, it surpassed the level of AlphaGo. Its performance was so outstanding that it was no longer worth competing against humans. Lee Sedol suddenly announced his retirement in 2019. He said that after facing an opponent he could never beat while playing against AlphaGo, Go lost its meaning for him.

However, since then, Google DeepMind has shown no clear achievements. Having conquered Go, DeepMind formed a separate team called ‘DeepMind Energy’ to pursue power efficiency in collaboration with the UK’s state-owned National Grid, but this project was canceled last year and the team was disbanded. DeepMind’s technology seems to work well in limited environments like Go or chess but does not fit the complex real world.

IBM’s artificial intelligence ‘Watson’ attracted global attention in 2011. This was because it defeated two champions by a large margin on the famous American quiz show ‘Jeopardy.’ Tens of millions of viewers were shocked to see a machine created by humans surpass people. Encouraged by this, IBM announced plans to apply Watson to finance, law, and healthcare sectors. It was actually installed at cancer research centers in the United States.

However, the University of North Carolina and New York Memorial Sloan Kettering Cancer Center stopped developing Watson for cancer diagnosis, and Houston’s MD Anderson Cancer Center declared failure after investing huge research funds for four years. IBM classified Watson as a failed business and decided earlier this year to sell Watson Health, the division overseeing medical AI.

Zoom Pizza was a venture company selected as one of the best startups in 2017 for its smart process using AI and big data. However, unable to surpass the taste of human-made pizza, it decided to shut down its robot pizza delivery business in January 2020.

The Todai Robot Project was started in 2011 by Japan’s National Institute of Informatics team, aiming to pass the University of Tokyo entrance exam. The AI robot had to score highly on both multiple-choice and essay tests, but it ranked only in the top 20% among all students, failing to gain admission to the University of Tokyo.

Japan’s SoftBank launched the humanoid robot ‘Pepper’ in 2014 as the world’s first emotion-recognition robot. Pepper has provided services at SoftBank stores, cafes, banks, hotels, and hospitals. While Pepper was thought to be a robot that would threaten jobs, it was actually Pepper that was laid off. Adidas’s ambitious Speedfactory, which produced sneakers, was also discontinued after four years.

As can be seen from the above cases, although artificial intelligence has tremendous potential, not everything can be achieved overnight. AI approaches us quietly and gradually. Only by abandoning unconditional illusions about AI and preparing meticulously will it finally become a precious companion to humanity.

This content was produced with the assistance of AI translation services.