The very first bill regulating the field of AI is to assess the risks of various artificial intelligence systems and classify them accordingly. The most dangerous developments that pose a threat to people should be banned in the EU. European lawmakers have banned the use of AI for “cognitive-behavioral manipulation of people or certain vulnerable groups”. As such, the authors of the project recognize, for example, voice-activated toys that provoke dangerous actions in children. In addition, the European Union intends to ban automatic facial recognition in real time. They are prepared to make an exception only for the capture of particularly dangerous criminals, and only after obtaining judicial authorization.
The second category of systems with artificial intelligence involves those that present rather high risks, but for them the “window to Europe” will always be open. Before entering the local market, developers must register their offspring in the EU database so that officials can assess its capabilities. Strict control of high-risk “intelligent” systems will be maintained “throughout their life cycle”, explains on the website of the European Parliament.
The ChatGPT chatbot that made a splash fell into the third category – generative artificial intelligence. The chatbot and its “compatriots”, who may create text, images and other media products from World Wide Web data, will have to comply with all transparency requirements. First, it is necessary to inform users about the content that has been generated by AI. Second, developers will be required to control the digital mind so that it does not produce illegal “products”. Along the way, the creators of such systems must publish a selection of copyrighted data, on the basis of which artificial intelligence is trained. It is worth noting that the CEO of OpenAI, which created ChatGPT, Sam Altman did not appreciate the legislation of the Europeans. The entrepreneur fears that it will be too difficult to grow under the proposed bill. The company notes that the new rules could be so overwhelming that it will have to leave the European market. Recall that in May, Altman spoke at a Senate hearing and called on U.S. lawmakers to develop international standards to regulate AI.
For so-called limited risk systems, the draft law on artificial intelligence introduces minimum requirements. They must meet the basic principles of transparency, i.e. make users aware of the interaction with AI and label all machine-generated content. This, according to MEPs, will help citizens identify counterfeits on the web.
In addition to the classification of systems and sanctions, the authors of the project plan to give a unified definition of the concept of “artificial intelligence”.
Incidentally, the debate over the legislative regulation of AI in the European Union has been going on for years, but the rapid development in recent months of technology products such as ChatGPT has forced MEPs to speed up. The risks of negative impact of artificial intelligence on the labor market and society as a whole are increasing day by day.
From now on, the AI law will be submitted to the European Council for discussion. European Commissioner for the Internal Market Thierry Breton, in turn, urged his colleagues to hurry and find an agreement “in the coming months”. The final version of the document, limiting the dangerous development of artificial intelligence, should be prepared by the end of the year. However, according to AFP, the law will not come into force until 2026 at the earliest. But on the other side of the ocean, they are in no rush to fight the coming “rebellion of the machines”. According to the Washington Post, U.S. lawmakers plan to do “specific things” no earlier than the fall.
In the meantime
British musician Paul McCartney plans to release the latest Beatles song using artificial intelligence. AI helped the legendary rock band’s ex-soloist record a track with late bandmate John Lennon. The Guardian newspaper thinks it is the song “From time to time” (Now and Then), of which the singer recorded the demo version shortly before his death. The artificial intelligence program separated the musician’s voice from piano sounds and background noise, allowing McCartney to overdub the recording of his performance.
“We had John’s voice and the piano, and he was able to separate them using artificial intelligence. You say to the machine: “It’s the voice. Here is the guitar. Put away the guitar”, the artist shared his impressions of the “cooperation” with the spirit of the machine. Moreover, this is not the first time that McCartney has used the “services” of the AI. , during a recent tour, he once sang a “duet” with Lennon using innovative technologies.
Read the Latest Science and Technology News Today on The Eastern Herald.
