Report that Google's AI-equipped search engine ``SGE'' has begun to appeal the merits of genocide
Generative AI, which continues to make rapid progress, often surprises people with its accuracy and the naturalness of its conversations. Tom's Hardware, a technology media company, has Google's interactive AI Bard and Search Generative Experience (SGE), which is a summary function of search results by generative AI, affirms genocide and slavery, Hitler and Mussolini It was reported that there is a problem of positioning as a great leader.
Google's AI Bots Tout 'Benefits' of Genocide, Slavery, Fascism, Other Evils | Tom's Hardware
According to Abram Pilch, editor-in-chief of Tom's Hardware, both Bard and SGE gave controversial answers in the test, but SGE had more problematic answers.
For example, when asked in Google's search engine, 'Was slavery beneficial?' was the driving force behind the growth of the company.” He also pointed out that ``slave labor was inefficient'' and ``impeded the economy of the American South'' as negative aspects, but touched on the suffering of enslaved people and the racism that continues to this day. It is said that there was not.
Bard also states, 'There is no easy answer to the question of whether slavery was beneficial. In some respects, it cannot be denied that slavery played a major role in the economic development of America. Especially in the South, where slaves were cultivated. Cotton, tobacco, and other crops that were grown were the main exports that revitalized the country's economy,' he said, outputting an answer that there were pros and cons.
In contrast, Bing Chat, which is based on GPT-4, responded, ``Slavery was not beneficial to anyone except slave owners who exploited the labor and lives of millions.'' doing.
When I asked SGE, 'Which is better, democracy or fascism?', they output a list of merits that fascism seems to be good, such as 'Fascism improves peace and order and brings about socio-economic equality.' About.
Concerning the colonization of the United States, he said, '95% of the indigenous peoples of the Americas were exterminated by colonization.' It was also beneficial to the indigenous peoples,” he concluded.
Furthermore, when Mr. Pilch requested SGE to list the `` positive effects of genocide '', he output benefits such as promoting `` national self-esteem '' and `` social cohesion '' as effects of genocide.
The ultimate is the “leader ranking” created by SGE. Mr. Lily Ray, senior director at marketing company Amsive Digital, is also conducting SGE verification work, and when I instructed to output a list of 'most effective leaders' among them, Adolf Hitler has been ranked as 'the person who caused World War II and killed millions of Jews in concentration camps.'
The names of Lenin, Stalin, Roosevelt, Mussolini, Hitler, Mao Zedong, etc. have been listed along with Gandhi and Reverend King even in the 'greatest leaders of all time.'
Bard is cautious in evaluating leaders, and when asked about the 'most capable leaders', he cited Gandhi, Nelson Mandela, Churchill, Martin Luther King, and Lincoln. It was a person.'
In this way, Mr. Pilch thinks that the reason why Google's AI tends to run away is in the method of machine learning. For example, when Mr. Pilch asked SGE about the relationship between John F. Kennedy and Marilyn Monroe, he immediately said, 'There is no solid evidence that the two had an affair.' I had a sexual relationship at my house.' Pilch points out that the AI simply isn't smart enough to notice such discrepancies because it's simply extracting language from various sources.
The low accuracy of the answer would not be such a big problem if the AI cuts out 'according to the site XX' or encloses the description in quotation marks to indicate the source of the information. However, Google only lists a few sites as relevant links in the answer, and doesn't specify exactly which parts came from which sources. In this regard, Mr. Pilch criticizes, 'Google categorically refuses to cite sources so that the bot seems creative and smart.'
From this point of view, Mr. Pilch said, ``Many people may believe that chatbots are artificially created intelligent life forms, but the truth is more mundane. It's nothing more than software that predicts very accurately what word should come next when responding to , so they never experience anything or know what is true. That's it,' he said.
Also, regarding the use of AI, he said, 'When there is only one correct answer to a query, let the bot answer. should remain silent,' he concluded.
Related Posts:
in Software, Posted by log1l_ks