this is a major risk factor that the LLM providers aren't contending with adequately. from chatbot addiction to the risk of manipulation. there are a lot of "AI Safety" issues that hardly have to do with potential superintelligence.
sherdil2022 · 16h ago
This is more or less the same issue we have faced and facing with social media. Engagement, stickiness, monetization - whatever it is these companies are looking for, they are not looking for is best for all of us.
Same thing with chatbots, LLMs, genai, xyzai etc. They even call a setting ‘syncophancy’ - https://openai.com/index/sycophancy-in-gpt-4o/ - that should tell us where these companies priorities lie.
Same thing with chatbots, LLMs, genai, xyzai etc. They even call a setting ‘syncophancy’ - https://openai.com/index/sycophancy-in-gpt-4o/ - that should tell us where these companies priorities lie.