Half wayish through the seminar 2

mon22ndsept/mon22ndsept.md

@@ -0,0 +1,11 @@
+Ethical Dilemmas of AI Humor
+
+Team 10 looks at AI generated humor. In particular they discuss what concerns we should have when thinking about this application. There are many concerns they bring up but I will highlight some that I feel are the most nascent.
+The feedback loop of AI systems and personalized humor systems, the feedback loop from regular AI humor exposure allows many of troubling mechanisms (reinforcing stereotypes etc) to have a run away effect.
+The challenge of who to assign blame or credit is difficult, I feel that for many modern systems the negative consequences of the systems aught to be laid on the person closest to distributions. I.e it is the person who shared the offensive joke written by OpenAI, it is the Twitch Channel that let the AI wild without enough controls. As the legal and regulatory framework adjusts to the new technologies there is great opportunity for it to setup good or poor incentives for safe development.
+Lastly I think the regulatory challenges or policing AI and content creation is a fine edge of doing it properly and correctly to pushing your own agenda. Someone needs to decide what is offensive and no one can really do that but ourselves therefore it is prudent that we not let a handful of companies decide what we globally find acceptable or not.
+
+Question that I have are:
+- In the research of AI bot generating humor are they all transformer LLM based?
+- What say you on the Seinfeld situation, should it of been cancelled?
+- Did any of the papers test to see how differentiable the AI and human generated humor was?