The fact that an AI-generated image was able to bypass the censorship of platforms like Instagram reminds us of the importance of “deviant” features that don’t quite work as expected, and can sometimes bring out positive things.
Sometimes the Numerama Rule 30 newsletter comes at bad timing. As I wrote the last edition, we were just beginning to question the impressive virality of All Eyes on Rafah, an image generated by artificial intelligence to protest the Israeli army’s attacks on a refugee camp. are in Palestine. A week later, the matter has already been dissected in detail.
We know that it was created by a participant in a Facebook group dedicated to promoting AI in Malaysiathen shared by another Internet user on Instagram (who denies having stolen it). We understood that its virality was linked to its unreal and smooth appearance, which allowed it to escape the moderation of the social network. We wondered if it was effective to express our solidarity with an ultimately very empty image.
Many people also defended its sharing, denouncing unfair contempt towards Internet users who certainly had good intentions. We tend to quickly judge activism on social networks, especially when it touches on subjects that annoy dominant people (anti-racism, sexism, etc.). However, we can recognize that the popularity of this image is astonishing. Why did it resonate so much with non-politicized people, stars who had not communicated anything about the conflict until now, artists or journalists, usually very vocal about the negative effects of generative AI? about their job?
Instagram is invaded by AI images on Palestine
I am sometimes asked to cite examples of feminist artificial intelligence. This question bothers me, because it assumes that AI (a marketing term in which we can put very diverse things) are both the problem and the solution. There is something about that in the defense of the All Eyes On Rafah image. Since we are told that AIs are inevitable, as much as they serve a good cause. This attitude is comparable to our love-hate relationship with recommendation algorithms. We know that they are biased, incomprehensible, that they serve ultimately economic interests, not the common good. We still hope that they will be on our side if we train them, if we replace certain words with emojis, if we take a selfie before sharing a story that is too political. We can’t imagine a world in which we could communicate without them.
On the one hand, these behaviors are well-intentioned. We do what we can with what we have. On the other hand, without popular reflection on the ethics of AI, we will only reinforce a media system that favors bad actors much more than good ones (which leads to another complex discussion about who can be placed in the ” good » actors). Instagram is now invaded by artificial images dedicated to the situation in Palestine, some created by sincere people, many others looking for virality. The rare photos of the violence, already harshly censored by the moderation of the social network, find themselves drowned out.
The success of the All Eyes On Rafah image is a bug. She made the machine work, for a brief moment, differently. This is an opportunity to ask ourselves essential questions: why must we negotiate with social networks so that they achieve their stated goal, putting us in contact with each other? Rather than adapting to this inevitability, how can we imagine other communication tools?
“ It is when the machine is out of service that we can look at it closely, dissect it, understand its workings and question ourselves about the meaning of our use. », explained the philosopher Marcello Vitali-Rosati, recently interviewed by France Culture on his essay In Praise of the Bug (La Découverte editions, 2024). Debating how social networks work is not the priority in these dark times, but I don’t think it’s off topic either. We can be relieved that the general public cares about such a dramatic event, for example. all means. We can also criticize this media system which reduces information to a pale copy of reality.
Subscribe for free to Artificielles, our newsletter on AI, designed by AIs, verified by Numerama!