Shulou(Shulou.com)11/24 Report--

Image source: the halo of PexelsChatGPT has been too shiny.

Since its inception, ChatGPT has been in the limelight: a new AI chat robot tool launched by OpenAI, which can quickly generate articles, stories, lyrics and even code according to users' requirements, answer questions, and so on.

At the time of its launch, it encountered people's viral spread because of the amazing amount of information and completeness in the answer, and had millions of users overnight. It also helped OpenAI secure a new $10 billion investment from Microsoft, bringing OpenAI's latest valuation to $29 billion. You know, when Google bought DeepMind, it cost only $600 million.

With the development of more than a month, ChatGPT seems to have gone out of the stage of being "flirted" by users and began to really show its potential. And similar AI tools are beginning to be really used by the industry.

| News website: warmly welcome these two days, one of the most discussed news in Silicon Valley is that Buzzfeed, a new media site based on the halo of ChatGPT and even OpenAI, has made a comeback and its share price has directly tripled!

The reason is simply that Buzzfeed announced that it will use the artificial intelligence API-- provided by OpenAI, not even ChatGPT itself, which has been misrepresented by some media, to help create some content.

Jonah Peretti Peretti, chief executive of BuzzFeed, said in a memo: "by 2023, you will see us transform the content of artificial intelligence that is still in development into part of our core business, thereby enhancing the Quiz experience, informing our mental ideas and providing personalized content to our audience."

Compared with regular news sites, Buzzfeed for young people is known for a variety of online tests, including "testing which princess you are in Disney" and "which superhero in the Avengers is the best boyfriend for you".

Its cooperation with OpenAI this time will be mainly applied to the production of this kind of "fast food" content. Specifically, BuzzFeed will use OpenAI's artificial intelligence technology to help generate relevant test questions on the site, thus helping brainstorming editors to find better ideas.

"to be clear, we are seeing the breakthrough in artificial intelligence usher in a new era of creativity, which will enable humans to use creativity in new ways to create unlimited opportunities and applications," Peretti said. "in terms of publishing, artificial intelligence can benefit content creators and audiences, inspire new ideas and invite audience members to work together to create personalized content."

Whether readers are really willing to pay for the fun quiz created by AI, the news of this collaboration is enough to bring BuzzFeed back to life.

BuzzFeed's shares have fallen more than 90 per cent since its listing through SPAC in December 2021, widened its third-quarter net loss to $27 million from $3.6 million a year ago, and even had to cut about 12 per cent of its staff to rein in costs. But as soon as the news of holding hands with OpenAI came out, its share price soared by more than 300%.

The next collaboration between BuzzFeed and Meta may bring the content generated by artificial intelligence to a wider range of users. Not long ago, Meta paid BuzzFeed millions of dollars to get BuzzFeed to generate content for Meta's platform and train creators on the platform. This also means that, next, you may be able to play a lot of mindless quizzes generated by artificial intelligence on Facebook and Instagram.

However, a spokesman said BuzzFeed will not currently use artificial intelligence to help write news reports. This decision may have something to do with the fact that another media used artificial intelligence to create content but was turned over not long ago.

In the application of artificial intelligence to news writing, CNET has gone further, but also ate the bitter fruit earlier.

As part of the CNET Money team's "testing" project, the editorial department has been using the internally developed AI engine since November 2022, generating 77 news stories, accounting for about 1 per cent of the site's total articles, according to CNET. The articles are uniformly signed "CNET Money Staff" to help editors create "a basic interpretation model" around the theme of financial services. The articles, written using AI tools, include, "will home equity loans affect private mortgage insurance?" And "how to close a bank account".

"the editors first generate an outline for the story, then expand, add, and edit AI drafts before publishing." Connie Guglielmo, editor-in-chief of CNET, wrote.

Soon, however, the CNET Money editorial team discovered that one of the articles was inaccurate. So they conducted a comprehensive audit. The result of the review was that a small number of articles generated by AI needed a lot of correction, while others had some minor problems, such as incomplete company names, ambiguous language or numerical errors.

For example, in an article "what is compound interest?" At the end of the article, AI gives some very inaccurate personal finance advice. "the early version of this article advised depositors to deposit $10000 in a savings account and earn 3 per cent compound interest a year, so that they could earn $10300 a year later." In fact, anyone who has studied math in primary school knows that savers can only earn $300.

Guglielmo did not say how many of the 77 published reports needed to be corrected, nor did it specify how many "substantive" problems and "minor problems" there were, except that corrections were listed below these articles.

However, because more than half of the reports contained factual errors or misquotations, CNET now stops using the artificial intelligence engine.

In fact, using AI to automate news reporting is not new. The Associated Press began to do so nearly a decade ago, but with the rise of ChatGPT, the issue has received new attention. How much specious content is mixed in when AI is used in content production on a large scale?

Despite these problems, Guglielmo opened the door to a return to the use of AI tools, saying that it would resume using artificial intelligence news writing tools after the problem was resolved.

| Education and academia: although challenges have begun to be boldly applied in the press, AI tools like Chatgpt have been questioned in more writing scenarios. These include the most popular but questionable place-the school.

To test ChatGPT's ability to generate answers in four-course exams, professors at the University of Minnesota Law School recently asked ChatGPT to take the exam and blindly judged the results. After completing 95 multiple-choice questions and 12 essay questions, ChatGPT got a ClearMurray average-a low but passing grade in all four courses, "flying low over the pass line".

On the Wharton Business School's business management exam, the ChatGPT did even better, scoring from B to B -. Christian Terwiesch, a Wharton professor, says ChatGPT is "very good" at answering basic operational management and process analysis questions, but does poorly in dealing with more advanced tips and has made "surprising mistakes" in basic math, some of which are even at the level of primary school math.

What does that mean? If left unchecked, ChatGPT will become the most powerful cheating tool in history-helping students with their homework or even completing exam papers.

Therefore, as the test results are released, more and more schools and teachers have expressed concern about ChatGPT's ability to cheat. For example, public schools in New York City and Seattle have banned students and teachers from using ChatGPT on school district networks and devices.

Professor Terwiesch also said he agreed that restrictions should be imposed on students taking exams. "A ban is necessary," he said. "after all, when you award a doctor's degree, you want them to really have medical knowledge, not just how to use chatbots. The same applies to other skills certifications, including law and business majors."

But Terwiesch believes the technology will eventually show up in the classroom. "if we end up with the same education system as before, then we're wasting the great opportunities offered by ChatGPT." He said.

In academia, ChatGPT has encountered harsher scrutiny.

Holden Thorpe, editor-in-chief of Science, a major US journal, announced an updated editorial policy banning the use of text from ChatGPT and saying that ChatGPT could not be listed as a collaborator.

Scientific journals require authors to sign a statement promising responsibility for their articles, Holdonso said. "but because ChatGPT can't do that, it can't be an author." He believes that even in the preparation of the paper, the use of ChatGPT is problematic. "ChatGPT has made a lot of mistakes, which may go into the literature." He said.

Not only Science magazine, but also other publishers have made similar moves. Springer-Nature, which publishes nearly 3000 journals, also issued a statement saying that ChatGPT cannot be listed as an author.

Perhaps the most stringent is the online programming question and answer platform Stack Overflow. Shortly after the launch of ChatGPT, it announced a total ban on responses from ChatGPT and any non-artificial responses, and further banned users from violating them.

This article comes from the official account of Wechat: Silicon Man (ID:guixingren123), author: VickyXiao

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.