5 Comments

1. 

   Anne Marte Gardenier on 2024-07-03 at 11:34

   Dear Guilherme Palumbo,

   Thank you for your presentation. I think the topic of ethical guidelines in AI is a very important and relevant issue. I have a clarificatory question: could you explain a bit more about how you prompted the LLMs to rate the ethics principles in different domains? What exactly did you ask them, and what kind of responses did you receive?
   I am interested to know more about this (and possibly due to my non-technical background, I currently lack the understanding).

   Thank you.

   Reply
2. 

   Roxanne van der Puil on 2024-07-03 at 11:35

   Dear Guilherme Palumbo, Davide Carneiro and Victor Alves,

   Thank you very much for presenting the results of your research. I do not have a technical background, but find the results hugely fascinating. A first clarificatory question I have is, what do you mean when you say LLMs are agnostic? Second, could you say more about why the SD score says something about the reliability of an LLM, and whether this is good or bad? Earlier in the presentation you mention that the AI experts showed a greater range of responses which, perhaps indicates nuance (a good thing in my opinion). Would we not want LLMs to show the same nuance? Therefore, could a higher SD be good/positive?

   Reply
3. 

   Anne Marte Gardenier on 2024-07-03 at 11:35

   Dear Guilherme Palumbo,

   Thank you for your presentation. I think the topic of ethical guidelines in AI is a very important and relevant issue. I have a clarificatory question: could you explain a bit more about how you prompted the LLMs to rate the ethics principles in different domains? What exactly did you ask them, and what kind of responses did you receive?
   I am interested to know more about this (and possibly due to my non-technical background, I currently lack the understanding).

   Thank you.

   Reply
4. 

   Roxanne van der Puil on 2024-07-03 at 11:36

   Dear Guilherme Palumbo, Davide Carneiro and Victor Alves,

   Thank you very much for presenting the results of your research. I do not have a technical background, but find the results hugely fascinating. A first clarificatory question I have is, what do you mean when you say LLMs are agnostic? Second, could you say more about why the SD score says something about the reliability of an LLM, and whether this is good or bad? Earlier in the presentation you mention that the AI experts showed a greater range of responses which, perhaps indicates nuance (a good thing in my opinion). Would we not want LLMs to show the same nuance? Therefore, could a higher SD be good/positive?

   Reply
5. 

   Taha Duri on 2024-07-06 at 13:35

   Thank you for the excellent research in this, still, under examined area. I look forward to future work, maybe, regarding the sociological analysis that may come out of this thorough comparative study where the human element is recognized as the guardian of the ethical use of AI, rather than considering AI a competitive or a threat to the human agency (which I, by the way, consider it not to be.)

   Thank you again.

   Reply