Existential safety

Continuing AI capailities advancments is catastrophic primarily not because of the biases the models might have or jobs people might lose but because, as half of ML researchers believe, there's at least a 10% chance of AI-induced existential catastrophe. We have to ensure the goals of first highly capable AI align with human values, but

источник: aiimpacts.org/how-bad-a-future-do-ml-researchers-expect/

theft

They are funding their research with stolen data from ordinary people and think they are being clever about it. We know you are deeply entangled with Stability AI. You are terrible people.

What training data would this use? Without knowing that, we'd be potentially funding the development of a powerful model that can be used against our interests. It's quite disturbing that this doesn't seem to be covered extensively in the proposal.

EXISTENTIAL RISK Do we want to gives to every human the capability to have intelligence at home in order to engineer potentially lethal patents like proven in the NATURE article from last year ? No. Do we want to accelerate research on a paradigm creating unreliable, unpredictible, and gazlighting machines, litteral Black boxes ? No. Do we want to accelerate Research on creating more and more powerful systems while we have no idea of how to steer a Super Intelligent AI ? No.

источник: Basic Argumentation

Existential safety

Continuing to advance AI capabilities is catastrophic primarily not because of the biases the models might have or jobs people might lose but because, as half of ML researchers believe, there's at least a 10% chance of AI-induced existential catastrophe. We have to ensure the goals of first highly capable AI align with human values, but

источник: aiimpacts.org/how-bad-a-future-do-ml-researchers-expect/

Oppose, unethical AI, grifters, scam

Since Laion datasets are created through scraping private, intelectual and copyrighted property continuing their development should be stopped and regulated. Further asking for open source AI created out of copyrighted material equals to stealing that many artists already complained about and hopefully will also start lawsuit against this as well.

источник: CAA lawsuit, Getty images lawsuit, Shutterstock lawsuit, Glaze project, Hive project.

Wouldn't this be used by rogue states?

#illegal, #open source, #copyrighted,#Egair, #Lawsuit

Oppose, unethical AI, stealing datas from artists, illegally used of fair-use. Since Laion datasets are created through a scrape of the web, with any consideration on intelectual and © property, continuing their development should be stopped and regulated. Advocacy for open source AI created out of © material is an illegall behaviour an act of stealing the work of so many alive artists so that why I hope that we, artists, will have the possibility to start lawsuit against Liaon as well.

источник: Source: CAA lawsuit and and Stable Diffusion litigation lawsuit

The petition assumes that the current, extremely expensive, large language models are the most promising direction in AI research.While these models have many interesting properties, their lack of truthfulness makes them unfit for most applications. We do need a computing facility of the kind proposed, but this petition over-sells one specific AI approach. We need greater diversity of approaches.

источник: Thomas Dietterich, Oregon, USA

collecting children abuse for collection, keeping exploitative images and links toward children, private children photos, abuse of children, allowing creation of realisting imagery of children abuse

Laion collects and keeps following data: Child abuse, Child pornography, private medical images that were uploaded and transfered between doctor and private persons, abuse of women, non consentual use of private information. All of this was proven in every data set so far. And was as such reported to multiple legal teams.

источник: laion-aesthetic.datasette.io/laion-aesthetic-6pls

How to trust the requests of a company whose data use permissions were limited by law to research uses only and that under a clear conflict of interest granted said data to companies with commercial interests? Those who were the first to use data that did not belong to them cannot be specifically trusted by giving it to companies with commercial interests