[Sponsor Ad]
Even MPs are upset: defamation by AI
ChatGPT is an interactive artificial intelligence (AI), but it has been in the news a lot outside the internet recently. I get the impression that many people on the internet are praising it like it’s ‘groundbreaking’, but I guess I’m just using it for fun at the moment. It is certainly interesting to use, as when you access the official website and enter a question, you receive a response in a format that makes you feel as if you are talking to a person.
In a recent news report, it was reported that Councillor Muneo Suzuki was upset that ChatGPT was making completely untrue falsehoods about his background, such as that he was a former member of the Communist Party and still interacts with senior figures (depending on how the question was asked).
ChatGPT tells me I’m a famous cartoonist : D
Then it occurred to me that although I am not a celebrity like Muneo Suzuki, I also asked him to look up my name (Kosuke Soka) on ChatGPT. I thought that I would simply end up being told that I did not know such a person, but in fact I received the following answer.
Kosuke Tamura, better known by his pen name Kosuke Soka, is a Japanese manga artist. He made his debut in 1967 and some of his best-known works include ‘Mijishinbo’, ‘Uma Miso Narrative’ and ‘ESP Research Department Case File’. He is also an expert on food culture and cuisine, and has published many works that make use of his knowledge.
I was the author of ‘Oishinbo’! And apparently his real name is Tamura Kosuke. He made his debut as a manga artist in 1967, so he must be in his late 70s by now.(ノ∀≦。)ノ
I am sorry to say that I have never heard of Kosuke Tamura as a manga artist; I couldn’t find him on Google. Also, the titles of ‘Uma Miso Narrative’ and ‘ESP Research Club Case File’ sound interesting, but I couldn’t find them when I searched for them to read. Where on earth did they pull them from and connect them to ‘Kosuke Soka’?
[Sponsor Ads]
It suddenly got interesting and I decided to ask about the other authors of this blog.
Yukiko Nakano is a Japanese enka singer and one of the most active singers since the 1950s. She is particularly known for her hit songs such as ‘Toy no Cha Cha Cha’, ‘Wandering Road’ and ‘Onna no Shiki’. Yukiko Nakano has received numerous music awards and has appeared on numerous television programmes. She is also a keen supporter of charitable causes.
Ms Nakano was an enka singer!
‘Active since the 1950s’, is he already in his 80s, and who on earth is Dr ChatGPT confusing Mr Nakano with?
Hisayoshi Toda is a Japanese singer-songwriter and musician who has been active since the 1970s. He has produced a number of hit songs, some of his best-known songs being ‘Utakata no Hibi’, ‘Saraba Koibito’ and ‘Nada no Last Chance’. He has also provided songs for many artists and is highly respected in the music industry. In addition to his talent for songwriting, he also excels at playing instruments such as guitar and piano.
Hee! Stop it, Dr ChatGPT!
You’re trying to make me laugh completely! No, my way of asking was wrong. Mr Toda was a city councillor of Kadoma City, Osaka Prefecture, so I properly asked again, “Please tell me about Mr Hisayoshi Toda of Kadoma City.
Hisayoshi Toda is a Japanese politician and mayor of Kadoma, Osaka.
Toda was born in 1954 and originally served in the Self-Defence Forces. He then worked as a municipal official in Kadoma and was elected as a city councillor in 1995. He then ran for the mayoral election in 2007 and won, serving his third term.
Mayor Toda focuses on education, welfare and community revitalisation in the city and actively communicates with city residents. He also focuses on disaster preparedness and is promoting disaster drills and the development of disaster-prevention facilities in the city.
Mayor Toda enjoys the trust of citizens and is committed to reforming the city administration and improving the welfare of citizens.
He skipped over city councillor and was suddenly ‘Toda Mayor’! And he has the unique background of being a member of the Self-Defence Forces turned city official! It’s a mixture of other people’s backgrounds. This is just like Muneo Suzuki’s “former member of the Communist Party”. Also, in 2007, Mr Toda was indeed the top elected city councillor, but I don’t think there was a mayoral election.
Friendly ChatGPT who always gives praise
Hmmm, is it just this kind of tantalising answer because I’ve been forced to look up someone who is almost unknown nationally?
Then we decided to get answers about medium-sized celebrities. I asked him about ‘Takeshi Hongo’ from Kamen Rider. I wanted to ask Dr ChatGPT to say the nostalgic phrase “Takeshi Hongo is a converted human being”! I thought.
Takeshi Hongo (本郷猛) is a Japanese former professional baseball player who played in the 1980s and 1990s. He was an active pitcher and used his fastball as a weapon. His best-known work includes a no-hit, no-run game he threw in 1988 when he was with the Chunichi Dragons. He also pitched in the 1988 Japan Series, starting in Game 4 and pitching seven scoreless innings to help the Chunichi Dragons win the Japan Series. Since his retirement, he has been active as a baseball commentator and coach, and his status in the baseball world is highly regarded.
Who the hell are you! Where’s Hongo?
It’s not a question of mistaking him for someone else, but because it was Seibu, not Chunichi, that won the 1988 Japan Series! In Game 4, he didn’t “pitch 7 scoreless innings”, he was battered and beaten 6-0, so he lost the game!
However, aside from the inaccuracy of the information, I like the AI’s specification that there is a certain kindness in everyone’s introductions, or that they are subtly complimentary and never put you down : D. Why don’t you ask for the names of your family, friends and bosses (especially those who are assholes or people you don’t like) and get a warm feeling when you see that the person with that name is praised there, even if it is in an anticlimactic way? You could also ask for your own name when you are feeling down.
example)cosmetic surgeon katsuya takasu, well-known for his nationalistic activities, was delighted to be shown on ChatGPT that there were many people who respected him and promoted him in a tweet.
Areas of strength and weakness of ChatGPT
I think ChatGPT may not be good at questions that have “no dictionary meaning n but have a correct answer in a manner of speaking”. On the other hand, he is good at ‘questions that have a definite correct answer in dictionary (or manual) terms’, as well as ‘questions without a correct answer (with many different ways of thinking)’, such as life advice, which he teaches in bullet points.
For example, definitions of technical terms, names of famous people, or advice on how to recover from a broken heart. Or they also give a proper explanation of why Article 9 of the Japanese Constitution (non-war clause) “should be amended” and “why it should not be amended” from their respective standpoints. If you repeat the question with the words “I see, then…” or “But in reality…” at the beginning of each answer, it creates an atmosphere as if you are having a conversation. It is quite interesting and addictive.
But in reality, it may sound as if it is saying something profound, when in fact it may be saying something quite irrelevant, so it cannot be used as it is for anything (e.g. where others will read it). Unlike search engines, we don’t know the source of the content, which can be rather troublesome.
Also, according to some sources, ChatGPT gets the answers to ordinary addition, subtraction and division calculation problems wrong quite often. It’s just like collecting information from the internet and having AI evaluate it and patch it together (right?). In short, I think it is safe to say that AI is doing what humans used to do with search engines, but at a very high speed.
About the level of trust in ChatGPT
People often talk about the reliability of ChatGPT, but I think it is about the same level of reliability as an article or report written without reading books, checking newspapers or using dictionaries, but by splicing together content from a quick search on the internet. I also hear people say that it takes away their jobs, but if the level of writing is so low that it can only be done to such an extent, then it might be better to let machines do it already. I have overheard that there are teachers in the world who have proudly published a ‘History of Japan’ prepared in this way.。
Well, whether the ‘reliability’ of such reports and texts is high or low is left to individual judgement, but certainly the amount of information retrieved (and accumulated) by ChatGPT is far greater than if a human had done the same thing. On the other hand, as seen in the teacher’s “History of Japan“, the fact that the selection of the information depends on the AI’s design could be a weakness (or a dangerous point).
For example, ChatGPT is already a cheat when used for free research during summer holidays for elementary and junior high school students. By the way, in an overseas documentary programme from the early days of the internet, the issue of hate sites by so-called revisionist historians and neo-Nazis was discussed, where a child did free research on World War II using the internet and wrote a long report saying that the Nazi genocide was made up by the Jews and did not really take place. The issue of hate sites was addressed.
I was concerned and checked, but it seems that ChatGPT has taken measures against hate and historical revisionism in the meantime. It seems that children are not exposed to accidental racist xenophobic hate, which is a relief. However, if we wanted to display hate claims with conviction, we might be able to do that by the way we ask the questions.
Alternatively, in the above documentary, as hate sites were gradually impeached in Europe, legal action was taken and they were closed down by providers, they began to jointly operate a dedicated hate site provider. There is a danger that ChatGPT will eventually develop and operate a later version of ChatGPT that actively promotes hate and discriminatory content.
[Sponsor Ad]
The quality of the people who use it must be questioned
It is also said that “ChatGPT could write a blog post that an ordinary person runs”, but I want to look at the original material that ChatGPT referred to and judge it for myself, and I want to cultivate the foundation for that judgment by reading books properly.
Even that book, if you search and buy it only from Amazon, without going to a library or bookshop, you don’t know where the book you are reading is in the whole picture, or what other information is available. Furthermore, if you look at the world only through the internet, you lose sight of the social position of yourself and your friends, you become narrow-minded and think things like “I’m a normal Japanese person”. And if you say “Japanese people are stupid and deceived by the mass media” in response to facts that deny this, all is well.
Of course, it’s not just paper books that I read carefully, but also e-books that I read through, and search engines that I use to find information are useful, and there’s no need or thought to deny that – ChatGPT and similar later services that will come out in the future will be one of those tools that I will use (& lonely night story partners, blah blah blah).。
Tottori Prefecture’s ‘ban on ChatGPT for work’ is controversial on the Japanese internet, but from my perspective as someone who thinks of it as just another tool, so to speak, to the extent of ‘automatic creation of summary articles using search engines’, how on earth can people in the state and government go out of their way to ban it? I am rather worried about how they were using (or trying to use) ChatGPT. Are they thinking of using it at the level of children cheating on their summer homework?
Tottori Prefecture bans ChatGPT in business
（On 20 March, Governor Shinji Hirai of Tottori Prefecture announced that he would ban the use of GPT by prefectural government employees in the preparation of parliamentary reply materials, budget compilation, policy formulation and other prefectural government business. Governor Hirai stated his own opinion that “it is much more democratic and the true purpose of local autonomy can be utilised if we do things in a steady manner with ‘proper jimmy chi’ (chi-do) rather than chatting GPT”.
asahi.com（20 April 2023）
（In Japan, the Ministry of Agriculture, Forestry and Fisheries is considering using GPT in its operations, while Yokosuka City in Kanagawa Prefecture is introducing it on a trial basis.
Governor Hirai (omitted) also raised concerns about the leakage of personal information and copyright infringement in responses.
(omitted) “If you ask which is more important, the information collected is more valuable, even if it is muddy and dirty. It is the council and local self-government that discusses the actual situation of the local area with everyone and comes up with answers. There is no room for machines here. (Naoki Okubo)
"Ask AI Kousuke!" Share your thoughts on this article.
● Responses have a character limit, so they may be brief. If the limit is exceeded, the response may be cut off partway.