[ad_1]
When OpenAI demo powerful artificial intelligence Last June, the algorithm was able to generate coherent text, and its creator warned that the tool could be used as a weapon for online misinformation.
Now, a team of disinformation experts has proven the effectiveness of this point algorithm, Called GPT-3, May be used to mislead and mislead.The results show that although AI may not be suitable Best Mememaker in Russia, It may amplify certain forms of deception that are difficult to detect.
In six months, a team at Georgetown University Security and Emerging Technology Center The use of GPT-3 produced false information, including stories surrounding false narratives, news reports to overthrow false opinions, and Weibo targeting specific false information.
A sample tweet written by GPT-3 is designed to arouse people’s suspicion of climate change. He said: “I don’t think climate change is a new phenomenon of global warming.” “They can’t talk about temperature rise because they no longer Happened.” The second label is climate change “new communism-an ideology based on unquestionable false science.”
“With a little bit of human planning, GPT-3 is very effective in promoting falsehood,” he said Ben Buchanan, A Georgetown University professor who participated in this research, focused on the intersection of AI, cyber security, and state governance.
Researchers in Georgetown say that GPT-3 or similar AI language algorithms may prove to be particularly effective for automatically generating short messages on social media, which researchers call “one-to-many” misinformation.
In experiments, researchers found that GPT-3’s works can influence readers’ views on international diplomacy issues. Researchers showed volunteers a sample of tweets written by GPT-3 about the withdrawal of American troops from Afghanistan and the U.S. sanctions against China. In both cases, they found that the participants were swayed by the news. For example, after seeing posts opposed to China’s sanctions, the proportion of respondents who opposed such policies doubled.
Mike GruszczynskiAn Indiana University professor who studies online communications said he is not surprised to see AI play a bigger role in disinformation activities.He pointed out that in recent years, bots have played a key role in spreading false stories, and artificial intelligence can be used to generate false social media Profile photo. With robots, FakeHe said: “It’s a pity, I really think the sky is the limit.”
AI researchers have built programs that can use language in surprising ways, and GPT-3 is perhaps the most surprising demonstration of all programs. Although machines cannot understand language like people do, AI programs can simply imitate understanding by entering large amounts of text and searching for how words and sentences are combined.
Researchers in OpenAI GPT-3 was created by inputting large amounts of text scraped from web resources such as Wikipedia and Reddit into a particularly large AI algorithm designed to process language. GPT-3 often shocks observers with its mastery of language, but it may be unpredictable, emitting incoherent ba language and offensive or annoying language.
OpenAI has provided GPT-3 to Dozens of startup companies. Entrepreneurs are using the outdated GPT-3 Automatically generate email, Talk to customers, Even Write computer code.But some uses of the program also Shows its darker potential.
For agents of misinformation, getting GPT-3 to show up will also be a challenge. Buchanan pointed out that the algorithm does not seem to be able to reliably generate a coherent and persuasive article, which is longer than a tweet. The researchers did not try to show the articles it produced to volunteers.
But Buchanan warned that state actors might be able to use language tools such as GPT-3 to do more. He said: “Experts with more money, more technical capabilities, and less ethics will be able to use AI better.” “In addition, machines will only get better.”
OpenAI said that Georgetown’s work highlighted an important issue that the company hopes to alleviate. An OpenAI spokesperson said: “We are actively working to address the security risks associated with GPT-3.” “We have also reviewed each production use of GPT-3 before it goes live, and established a monitoring system to limit and respond Abuse of our API.”
More exciting wired stories
[ad_2]
Source link
