My view is that most of LLMs are trained on data obtained illegally. Like with any product we buy, companies must disclose sources and manufacturing practices: Fully disclose your training data and "humans-in-the-loop", so that we users can make an informed (and ethical!) decision.

So is it actually #ethical to use such GenAI system for producing research that should follow the highest ethical standards? Do we all need this "faster" productivity, if we don't have the community resources to evaluate all the outputs aided by AI? www.nature.com/articles/d41...

Ethics dumping: one stage of training large language model is ”reinforcement learning from human feedback” which became exploitations of workers from the global south:www.theguardian.com/technology/2...

Ethics bluewashing: open source large language models are actually not open at alldl.acm.org/doi/pdf/10.1... (Figure 2)

Ethics shopping: justifying any unsustainable practices (e.g. towards the environment) in the search for AGI (Artificial General Intelligence, AI that matches or surpasses human cognitive capabilities) for the benefit of humankind (?). www.thedailybeast.com/ai-may-soon-...

Ethics shopping: large AI companies have scraped the internet under the principle of ”fair use” (Fair use is a doctrine in United States law that permits limited use of copyrighted material without having to first acquire permission from the copyright holder.)

GenAI comes with various risks and ethical concerns related to fairness, accountability, transparency, trustworthiness. While the ethical principles for AI can be similar to core principles of responsible conduct of research, there are also **ethical risks** related to the practices around AI