Experts are trying to make artificial intelligence forget what it has learned

PARIS - When Brian Hood discovered that GPT chat linked him to a criminal past, this Australian politician found himself facing a dilemma that engineers are working hard to solve: how to teach intelligence to erase mistakes.

The legal option represented by Brian Hood threatening last April to file a defamation lawsuit against OpenAI, the company that created GBT Chat, does not seem to be an appropriate solution, nor does the solution lie in completely resetting the parameters of artificial intelligence, as training the model again takes a long time and is expensive.

Specialists believe that the issue of “unlearning,” meaning making artificial intelligence forget some of what it has been taught, will be extremely important in the coming years, especially in light of European data protection legislation.

Lisa Giffen, professor of information sciences at RMIT University in Melbourne, confirms that “the ability to erase data in learning databases is a very important topic,” but she believes that a great effort is still required in this field given the current lack of knowledge. On how artificial intelligence works.

Specialists believe that the issue of unlearning will be extremely important in the coming years, especially in light of European data protection legislation

With the massive amount of data on which AI is trained, engineers are seeking solutions that allow greater specificity, removing false information from the knowledge of AI systems in order to stop its spread.

An expert researcher in this field, Miqdad Karmanji from the British University of Warwick, explained to Agence France-Presse that the issue has gained momentum during the last three or four years.

Google DeepMind, which specializes in artificial intelligence, worked to address this problem, as experts from the American company published last month with Cormanji an algorithm dedicated to erasing data in important linguistic models, such as the GPT Chat and Bard models from Google.

More than a thousand experts participated in a competition launched by the American company, working between July and September to improve methods of “unlearning” artificial intelligence.

The method used, similar to what other research has found in this field, is to introduce an algorithm that orders the artificial intelligence not to take some of the acquired information into account, and does not involve modifying the database.

Karmanji confirms that this process can be a “very important tool” to enable research tools to respond, for example, to deletion requests, in accordance with personal data protection rules.

He pointed out that the algorithm that was developed also proved effective in removing copyright-protected content or correcting some biases.

Lucan clarifies that he is not saying that this algorithm is “useless, uninteresting, or bad,” but rather that he believes that “there are other priorities.”

Professor of Artificial Intelligence at the University of Edinburgh, Michael Rovatsos, considers that “the technical solution is not an effective solution,” and explains that “unlearning” will not allow broader questions to be asked, such as how to collect data, who benefits from it, or who should be responsible for the damage caused by artificial intelligence. .

Although Brian Hood's case was dealt with, without explanation, after it received widespread media attention that resulted in the correction of the data processed by ChatGPT, he believes that the methods that should be used at the present time should remain manual.

“Users should check everything in cases where chatbots give wrong information,” says Brian Hood.

International artists open up to “artificial intelligence arts” in Morocco

Casablanca - Local and international artists are gathering in the city of Casablanca as part of the first session of the “Artificial Intelligence Arts” exhibition, which was launched on December 9, to display artistic works they have created with the help of artificial intelligence technologies.

The exhibition, which continues until December 31st and is organized at the American Arts Center, offers opportunities to explore the new possibilities available to artistic creativity thanks to artificial intelligence.

The organizers of the exhibition explained during a press interview to present this artistic event organized by the Ofoto Arts Lab under the auspices of the Ministry of Youth, Culture and Communication, that the exhibition combines contemporary art, including photography, video, digital art, advanced digital technologies and artificial intelligence, and also opens innovative horizons in the field of art. Creativity and artistic expression.

The exhibition, in its first session, combines contemporary art, advanced digital technologies, and artificial intelligence

The exhibition aims to “actively engage the audience, and transform it into an interactive participant, by enabling it to explore its curiosity while creating personal connections with the works through artificial intelligence,” according to the exhibition’s curator, Hamid Al-Akhdar.

Al-Akhdar revealed in a statement to the Maghreb Arab News Agency that the first session will bring together well-known local and international artists who integrate artificial intelligence into their artistic works.

The exhibition features the participation of a number of Moroccan artists, including, in particular, Christian Maamoun, Idriss Al-Karnachi, Walid Bandra, Mehdi Sefrioui, Maad Abu El-Hana, and Yasser Ramzy.

Photographer Mehdi Sefrioui, who is also curator of the exhibition, pointed out that “these artists present a series of creations that aim to rework the Moroccan imagination, far from clichés and reductive ideas transmitted from a Western vision of our identity.”

The exhibition includes famous Dutch artists such as Louise T. Boyle, Rodger Werkhoven, the independent creative director at OpenAI, and Arno Koenen, who is known for his large-scale works in public spaces.

Through the exhibition, the public will also be able to discover distinguished works of art under the title “The Seven Wonders of the World” by the French group Obvious, known for its innovative integration of artificial intelligence and augmented reality.

It is noteworthy that the Artificial Intelligence Arts Exhibition project was ranked first out of 331 projects by the Fine and Visual Arts Support Committee of the Ministry of Youth, Culture and Communication.

The exhibition aims to constitute a forum for thinking about the limitless possibilities of artificial intelligence in art and a platform for dialogue on its ethical, aesthetic and philosophical implications.

Last November, the city of Casablanca hosted international artists of various nationalities, within the framework of the twenty-ninth session of the International Festival of Video Art, under the slogan “From videotape to artificial intelligence.”

The works that were displayed in the art space of the American Art Center during the period from 7 to 11 last November bore the signatures of contemporary Moroccan, Syrian, Canadian, French and Spanish artists, who used innovative techniques to address current topics.

The festival also proposes artistic creations that are open to all forms of digital creativity, most of which are presented for the first time, and represent: dance, digital arts, virtual and augmented reality, interactive installations, audio-visual performances, mapping and robotics.