...and regards for the INHERENT AND INDISPUTABLE rights of humans.....which may be FURTHER AUGMENTED AND/OR specified by a given profession or group...

I think the definition of ethical choice is still a little bit confusing and could be worded slightly better. Perhaps something along these lines: a situation in the context of automated and intelligent systems in which one of several options must be chosen on the basis of a set of principles and rules regarding moral obligations and the rights of humans and nature. Such a choice is made using a shared moral framework or set of moral principles based upon the entity’s Code of Ethics.

Algorithm Ethics would contain each of the following disciplines OFTHE SOCIO-TECHNOLOGICAL AI ECOSYSTEM: Proportionality, Risk Analysis, Benchmarking, Application of Soft Law, Necessity, Diverse inputs, and Multistakeholder feedback, Reliability, Validity, KPI design.

I would either include "impact" or change the phases to the "AI cycle" so that we don't miss a phase 🙌. Also I would add both the words: Algorithms and Models. Algorithms by themselves are seldom wrong, but the models that they create are the one that could be bias or present other kinds of problems.

Ethics Officer in a tech company to me is like having a Physics Officer in a tech company, as if there is some kind of C-level need to prove that gravity exists and explain at the board level why up isn't down. I'm not opposed, but I wonder if having one is evidence of a crisis state, disaster handling. Also the list of disciplines seems overly long. Does "necessity" come to a conclusion that the other disciplines are less necessary at the top? I would aim for a max of five.

• The 'Ethical Choice' technical term seems unnecessary. We could just simplify: "Algorithm Ethics is the study of the moral complexities involved in the design, development, and operation phases in the context of...." We then say all EOs need to have studied Algorithm Ethics.

• Shouldn't Ethical Choice be an outcome of studying Algoritm Ethics, and not a necessary component of the discipline? Seems confusing.

"a discipline of applying Ethical Choice in the design, development, operation and maintenance phases *of algorithms* in the context of automated and intelligent systems." Algorithms are run through calls from the platform (AIS), the visualisation of their output has specific Ethical considerations and is produced by other elements of the AIS.

We have different definitions and distinctions. I don't believe in "ethical choice" for instance. It's crucial that there be hierarchies and dependencies on these terms. For instance...everything starts with justice - once an org can name and weight the normative values of justice they're optimizing for, then they're able to define their own values, establish principles, codes, and governance. More specifically, I'll send Ryan some definitions via slack b/c I can't enter enough info here

I agree it is a coherent definition with coherent underpinning. My comment is about Ethical Choice and is for future discussion: “Options to be made” is not intuitive language. “Options to be decided” is my suggested change.

We may want to include the word "testing" in the parts of the process, unless we consider that part of development. Testing ML algorithms should activate an "Ethical Choice". "Deployment" might also be something to consider adding.