On August 10, an article was published on the World Economic Forum (WEF) portal AT solution to online abuse? AI a plus human intelligence (The solution to online abuse? AI plus human intelligence).
The author of the article is the vice president of an Israeli company Active hedge “with a background in intelligence” Inbal Goldberger, proposes to combine an artificial intelligence network with human intelligence to combat the “dark world of online abuse” – with “extremism, disinformation and hate speech.”
Artificial intelligence (AI) of computer programs, according to Goldberger, often cannot understand the “verbal formulations of hate groups”, it has not achieved “fluency in terrorist languages and a subtle understanding of disinformation campaigns.”
“For example, there are reliable AI models for detecting nudity, but only a few can distinguish whether this nudity is part of a renaissance painting or a pornographic image. Likewise, most models are unable to tell if the knife shown in the video is being used to advertise a butcher’s equipment or for a violent assault. This gaping hole can be exploited by attackers.”
Moderators who remove “unwanted” content lack the strength and knowledge to blacklist all misinformation sites. “Unlike AI, human moderators and subject matter experts can detect the nuances of online abuse… This accuracy, however, is limited to the analyst’s particular area of expertise: a human moderator who is an expert on white supremacy in Europe need not be able to recognize malicious content in India or disinformation narratives in Kenya.”
Goldberger makes an application for global censorship: they say that moderators are at their best in Europe, but they have not yet reached India and Kenya. It is understood that the Indians, Africans, Arabs, as well as the Chinese and Russians themselves cannot be entrusted with control over information on the Internet.
In order for the fight against “offensive content” to become truly global, Goldberger proposes to combine the efforts of AI (threat scaling) with the mind of country scientists.
The world’s networks went wild: “The Klaus Schwab World Economic Forum proposes to automate the censorship of ‘hate speech’ and ‘disinformation’ with the help of AI fueled by ‘domain experts’.” Such a symbiosis of man and machine is called hybrid intelligence, but Goldberger avoids this term, since the field of hybrid intelligence is one of the main areas of research and development of American intelligence agencies through the Advanced Research Agency (IARPA) in the field of so-called preemptive intelligence (Proactive Iintelligence).
In 2017 IARPA launched the development of a “human-machine hybrid forecasting system (human–car hybrid forecasting system) issues such as disease outbreaks, stock market fluctuations, elections, interstate conflicts.” In particular, the project “Synergetic Foresight of Geopolitical Events” (SAGE) was headed by the head of the research group on machine intelligence and databases of the University of Southern California Aram Galstyan. Worked out “feedback between man and computer intelligence.” In establishing such feedback, Inbal Goldberger sees the prospects for global censorship.
Project SAGE is part of a larger program Hybrid Prediction Competition (HFC), overseen by the Office of the Director of National Intelligence (ODNI) USA. The HFC program aims to compare the performance of “various human-machine hybrids at predicting real-life events by fusing human judgment with information from sources such as traditional news reports, social media, and financial figures,” it said. ODNI.
The possibility of using hybrid intelligence for intelligence purposes in the United States was taken up for the reason that previous projects of this kind had failed. The FutureMap project failed. Crowdsourced Intelligence Prediction Project Failed; its accuracy turned out to be less than 48%, that is, worse than when tossing a coin (heads-tails).
The project of proactive intelligence with the help of hybrid intelligence has also failed. Elon Musk has now been transferred to this direction, who promptly reorganized himself from promises to send people to Mars and started arguing that people should merge with artificial intelligence in the near future.
According to CIA Director of Digital Technology Andrew Hellman, American proactive intelligence has learned to accurately predict outbreaks of instability, conflicts and terrorist attacks. in 3-5 days before they happen. However, “predictive capabilities with an event horizon of 3–5 days can only suit ordinary people who are going on a picnic next weekend and are interested in the weather. For proactive intelligence of social instability and conflicts, a forecasting horizon of at least a couple of weeks is desperately needed, but in a good way at 3–5 weeks“, – says the leading Russian THIS IS-analyst Sergey Karelov.
Against the background of failures in the use of hybrid intelligence, it seems that the decision was not to try to predict the future, but to design it, as Nobel laureate Dennis Gabor taught. That is to do social engineering.
Inbal Goldberger’s article on the WEF website caused a lot of noise. “The WEF plans to use artificial intelligence technology to combat ‘hate speech and disinformation’ and now anything that goes against their agenda will be considered disinformation… Let’s call it a global dictatorship in the making,” writes British journalist Sophie Corcoran.
“Inbal Goldberger is promoting AI as the king of censorship… It looks like she has completely merged with the Biden team,” they say on social networks.
“The group behind the Great Reset is advancing global censorship in the ‘AI plus human intelligence’ scheme. “The World Economic Forum (WEF) has proposed a new way to censor online content that requires a small group of experts to train artificial intelligence to spot ‘misinformation’ and offensive content,” read the headlines.
One way or another, attempts to introduce global censorship with the help of artificial intelligence have not been marked by practical success. So far, much ado about nothing.
If you notice a mistake in the text, highlight it and press Ctrl+Enter to send the information to the editor.