"Art 6: “… actively disseminate positive energy, and promote the application of algorithms to be good.”" It depends on who it is "positive" and "good" for. For

"Art 6: “… actively disseminate positive energy, and promote the application of algorithms to be good.”" It depends on who it is "positive" and "good" for. For the society, the government, or the user? I would like to see algorithms that focus on the user. For that you want to have algorithms that are understandable and controllable by the user. Unfortunately, this is not compatible with the algorithms employed for recommendations today. They use neural networks that are hard to understand not only to the users but to the developers of these systems. Look at the abuse prevention technique used by DALL-E 2 [1]: "Our content policy does not allow users to generate violent, adult, or political content, among other categories. We won’t generate images if our filters identify text prompts and image uploads that may violate our policies." I worry that the algorithm regulation will focus on forcing recommendation systems to use similar controls and then those who control the filters of what is "not good" will effectively control what is acceptable to learn and talk about. I'd like to propose an alternative - design the recommendation algorithm that gives control to the user to decide what is good for them. I'm building https://linklonk.com https://linklonk.com as an experiment in this area and the algorithm is detailed here: https://linklonk.com/item/3292763817660940288 https://linklonk.com/item/3292763817660940288 [1] - https://openai.com/dall-e-2/ https://openai.com/dall-e-2/