openai_chatgpt，chatGPT

人工智能聊天机器人ChatGPT最近又闹了个笑话，将日本推动人工智能应用的数字化担当大臣误认为是日本首相。

ChatGPT misidentified Japan’s digital transformation minister amid his push for artificial intelligence-powered solutions to help overcome labor shortages in the country.

日本数字化担当大臣河野太郎5月15日在接受彭博电视采访时说：“我问ChatGPT河野太郎是谁，它给出了错误的答案。”他接着说道：“所以（在使用ChatGPT的时候）要小心。”

“I asked ChatGPT who Kono Taro is and he came back with the wrong answer,” Kono said in an interview with Bloomberg Television broadcast on Monday. “So you need to be careful,” he added. Kono asked that his name be written in the Japanese style, with his family name first.

ChatGPT是一种最新推出的由人工智能技术驱动的自然语言处理工具。据河野太郎说，他向ChatGPT询问他自己是谁，ChatGPT误以为他是日本首相。

Kono sAId ChatGPT, a natural language processing tool driven by AI technology, identified him as “prime minister of Japan” when he entered a query about himself.

4月30日，在日本群马县高崎市举行的为期两天的七国集团数字与技术部长会议结束后，日本数字化担当大臣河野太郎出席新闻发布会。图源：视觉中国

据《南华早报》报道，尽管河野太郎在日本媒体的民意调查中经常被认为是最适合担任日本首相的人，他还是在2021年的日本自民党党魁选举中输给了日本前外相岸田文雄。

He often tops media polls on who is the most suitable person to be prime minister, but lost to former Japanese foreign minister Fumio Kishida in a runoff vote for the leadership of the ruling Liberal Democratic Party in 2021, the South China Morning Post reported.

岸田文雄于2021年10月4日当选日本第100任首相，并于当年11月10日再次当选日本首相。

Kishida was elected as Japan’s 100th prime minister on Oct 4, 2021 and was re-elected as the prime minister in a special parliament session on Nov 10 that year.

5月15日，日本首相岸田文雄在位于东京的首相官邸接受媒体采访。图源：视觉中国

ChatGPT误认日本数字化担当大臣是日本首相一事或许反映出其小语种语库资料不足。

The recent mistake made by ChatGPT may show a lack of minority language data sets, which is a barrier for training AI software.

据彭博新闻社报道，河野太郎表示，日本政府正在与微软和其他科技巨头讨论创建日语相关的数据集。他说：“与英语相比，小语种的数据集并不大，因此可能存在偏差。”

Kono said the Japanese government is discussing data set creation with Microsoft Corp and other providers of the technology, Bloomberg reported.

“The minority language data set compared to English is not big, so it could be skewed,” he said.

据日经亚洲报道，日本政府5月11日召开了讨论制定人工智能使用规则的“AI战略会议”的首次会议，意在最快下个月建立指引生成式AI发展的框架。

The Japanese government’s strategy council on artificial intelligence convened for the first time on Thursday, with an eye toward establishing a framework to guide the development of generative AI as soon as next month, Nikkei Asia reported.

据报道，岸田文雄在会上表示：“人工智能具有正向改变我们的经济社会的潜力，但是它也有风险，因此正确处理好这两个方面很重要。”

Kishida said at the meeting, “AI has the potential to change our economic society positively, but it also has risks, so it’s important to deal with both appropriately,” Nikkei Asia reported.

今年4月30日，七国集团数字与技术部长会议的参会者呼吁加快关于负责任使用和治理人工智能的讨论。据《日本时报》报道，自从美国人工智能研究公司OpenAI去年11月发布ChatGPT以后，越来越多的科技公司宣布它们正在开发生成式AI产品。于是，对人工智能技术进行治理就成了此次会议上的一项主要议题。

On April 30, digital and technology ministers from Group of Seven nations called for accelerating discussions on the responsible use and governance of artificial intelligence.

Governance over the technology became one of the main topics at the G7 meeting, as a growing number of tech firms announced that they are developing generative AI products after AI research and development company, OpenAI Inc, launched ChatGPT in November, the Japan Times reported.

4月30日，在日本群马县高崎市举行的为期两天的七国集团数字与技术部长会议结束后，日本数字化担当大臣河野太郎（右二）出席新闻发布会。图源：视觉中国

据彭博新闻社报道，软银集团旗下的电信子公司软银公司近日宣布，它正在加入打造ChatGPT版本的全球竞赛。该公司于今年3月成立了一个新实体，挑选了大约1000人来开发ChatGPT的日语版本。

SoftBank Group Corp’s mobile unit has announced it is joining a global race to build a version of ChatGPT, Bloomberg reported.

SoftBank Corp set up a new entity in March, choosing about 1,000 people to develop a Japanese-language version of OpenAI’s artificial-intelligence chat technology, Junichi Miyakawa, president and CEO at SoftBank Corp, said on Wednesday.

记者：江雪晴

编辑：左卓

来源：彭博电视 南华早报 日经亚洲 日本时报 彭博新闻社等

China Daily精读计划来了！

每天20分钟，

带你学英语，看世界！