VOA科学技术2023 研究人员称生成式人工智能存在风险

（单词翻译）

Researchers: Generative AI Might Make It Easier to Target Reporters

研究人员称生成式人工智能存在风险

The artificial intelligence (AI) tool ChatGPT was released last year. Since then, the possibility that artificial intelligence might take over the world has worried people more than ever.

人工智能工具ChatGPT于去年发布。从那时起，人工智能接管世界的可能性比以往任何时候都更让人们担忧。

A new report from New York University's Stern Center for Business and Human Rights identifies eight risks of generative AI. Some of those risks especially concern reporters and news media organizations.

纽约大学斯特恩商业与人权中心的一份新报告确定了生成式人工智能的八种风险。其中一些风险尤其涉及记者和新闻媒体机构。

Disinformation, computer attacks, privacy violations¹, and the weakening of news media are among the risks the team reports.

该团队报告的风险包括虚假信息、计算机攻击、侵犯隐私和削弱新闻媒体。

Stern Center assistant director Paul Barnett was a co-writer of the report. He told VOA that people are confused about what risks AI presents now and in the future.

斯特恩中心助理主任保罗·巴内特是该报告的合著者。他对美国之音表示，人们对人工智能现在和未来带来的风险感到困惑。

Barnett said: "We shouldn't get paralyzed by the question of, ‘Oh my God, will this technology lead to killer² robots that are going to destroy humanity?'"

巴内特说：“我们不应该被‘天哪，这项技术会导致毁灭人类的杀人机器吗？’这种问题搞得不知所措。”

The systems being released right now are not going to lead to the extreme danger in the future some worry about, explained Barrett explained. The report urges lawmakers to face some of the existing problems with AI.

巴雷特解释说，目前发布的系统不会导致一些人担心的未来极端危险。该报告敦促立法者正视人工智能存在的一些问题。

Among the biggest concerns with AI are the dangers it presents for reporters and activists³.

对人工智能最大的担忧之一是它给记者和活动人士带来的危险。

The report says AI makes it much easier to dox reporters online. Doxxing is when a person's private information, like their address, is posted publicly.

该报告称，人工智能使得“人肉”新闻记者变得更加容易。“人肉”是指一个人的私人信息，比如他们的地址，被公开发布。

Disinformation is another problem, as AI makes it easier to create propaganda. The report noted⁴ Russia's involvement in the 2016 U.S. presidential election. It said use of AI could have widened and deepened Russia's interference with the process.

虚假信息是另一个问题，因为人工智能使得政治宣传更为容易。该报告提到了俄罗斯卷入2016年美国总统大选。它表示，人工智能的使用可能会扩大并深化俄罗斯对这一进程的干涉。

Barrett said AI "is going to be a huge engine of efficiency, but it's also going to make much more efficient the production of disinformation."

巴雷特表示，人工智能“将成为一个巨大的效率引擎，但它也将使虚假信息的产生更加高效。”

Disinformation could also be dangerous for news reporters because it could lead the public to trust them less.

虚假信息对新闻记者来说也可能有危害，因为这可能会导致公众对他们的信任度降低。

And AI could worsen financial problems for news media groups. People are less likely to seek news reports, the researchers say, because they can seek answers from ChatGPT instead. The report says that could shrink traffic on news sites causing losses in their advertising⁵ revenue.

人工智能可能会加剧新闻媒体集团的财务问题。研究人员表示，人们不太可能寻求新闻报道，因为他们可以从ChatGPT寻求答案。该报告称，这可能会减少新闻网站的流量，导致广告收入损失。

However, AI could also be helpful for the news industry. The technology can examine data, fact-check sources, and produce headlines speedily.

然而，人工智能也可能对新闻行业有帮助。该技术可以检查数据、核实来源，并快速制作头条新闻。

The report urges the government to supervise AI companies more in the future.

该报告敦促政府在未来加强对人工智能公司的监管。

"Congress, regulators, the public – and the industry, for that matter, need to pay attention to the immediate⁶ potential risks," Barrett said.

巴雷特说：“就此而言，国会、监管机构、公众和行业都需要关注眼前的潜在风险。”