As an English version of “JQ International Review Vol.11” published in November 2017

According to Wikipedia, it is said that AI research became an academic discipline at a conference held on the campus of Dartmouth College in the summer of 1956. Many of the participants at the conference predicted that machines with the same level of intelligence as humans would emerge. And that was the beginning. At that time, it seemed that the United States and the British government were funding millions of dollars to realize this vision. Still, the result was disastrous, and in 1973 it was criticized by British physicist Professor James Wrighthill. By Parliamentary pressure, the US and UK governments have stopped funding in research with unknown goals related to artificial intelligence.

Also, even seven years later, including the Japanese government and companies, it seems that more than 50 billion yen has poured into AI research. However, it still falls short of expectations, and records tell that investment was frozen again at the end of the 1980s.

After such a transition, AI has evolved considerably and is now used in robots, automobiles, smartphones, etc. However, I guess no one would deny that the essence of AI is a method using statistics and probabilities. From quoting from webbu.jp/ai-now-and-future-2143, it is a method of predicting the probability of action by statistics throughout various data, so-called “Big Data”. “Big Data” are such as search keywords on the Web, transmission on SNS, voice, text, images and data that can be acquired from multiple sensors in the real world represented by GPS.

In other words, since statistical probabilities are derived from a large amount of data, it is possible to imitate actual human behaviour patterns to some extent as a result.  But what if you cannot predict human behaviour patterns? For instance, in a car, a sensor could detect the driver’s behaviour then it might be able to prompt a break by the movement, but, for people with dementia, there is no behaviour pattern, so It might not be able to detect it.

Rather, would not it be the real utilization of AI to be that predicting what is not in the behaviour pattern. In other words, it can be said that it is impossible to create new behaviour as long as it is based on statistics and probability. What we like to expect from AI is not like chanting as if you saw the world ahead of another dimension to the things created based on past areas, but, including AI scholars, computer makers, and their affiliated companies, how can you predict for what you can’t predict. What we like to expect from AI is not like chanting as if you saw the world ahead of another dimension to the things created based on past areas. Still, including AI scholars, computer makers, and their affiliated companies, it is that how can you predict from what you cannot predict.

Given these factors, I think that we can expect AI to be such as a predictive of danger. Still, unless the direction that AI initially aims for, that is, the theory that does not depend on statistics and probability, is discovered, I think the future of AI wouldn’t be promising. In that sense, I named it “Death agony of Artificial Intelligence” As Looking at the essence of today’s academic fields, there is an increasing number of immature researchers who think that the research is to spin a computer-based on phenomenology because there is almost no attitude to tackle basic theory. In this case, not only AI but also basic science itself would not take its root and might eventually be left out of the world.

We hope that this JQ International Review can further push and encourage to our readers.