AI and Regulations
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis essay places itself on an oft beaten path about how to conceptualise AI artefacts to make them amenable to a better comprehension on behalf of regulators and, hence, to a more effective regulation.
In this sense, it does not break new grounds, since many scholars have already extensively written on this theme, with alternating results (see Keith Miller, Deborah Johnson, Vincent Müller, Luciano Floridi etc etc).
Still, it is written with notable clarity, which makes it a pleasure to read, very understandable also to an audience of non-specialists, which makes it indubitably deserving of publication.
There are some conceptual issues, though, that the author should address to improve the essay. Here and there, they stress that one stark difference between human agents and artificial agents derives from the fact that the former conceive, design, and create the latter (as explicitly written in lines 181-183). However, in the rest of the essay this real agent vs made object duality is often forgotten, in favour of an anthropomorphisation of the latter, especially in the second half of Section 5.
The author must pick a side - either fully highlighting the artifactual nature of AI and hence giving up describing it in metaphorical terms (such as "knowing", "being prisoner" etc.) or embracing the metaphor. The latter choice makes for an easier reading, but I am afraid it may defeat the very purpose this essay was written for. Maybe the author can go for the second choice, with an additional paragraph that keeps the distinction between human and AI agents clear throughout the text.
Another point that needs improvement is the seemingly partial adoption by the author of the STS (Science, Technology and Society) paradigm when it comes to Machine Learning systems. See the case of LLMs (lines 111-112): society is involved not only when the system is deployed; society is heavily involved when it produces the data with which the system is trained. I am sure that the author is aware of this further connection between society and technology: they should only make it explicit in the text.
Minor typos:
lines 16-17: the abstract seems to be missing a part here
line 50: the title of Section 2 seems to be missing a ":"
line 479: the title of Section 7 seems to be missing a ":"
line 484: there should be a "." between "destiny" and "Rather"
Author Response
Please see attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper is a review of the understanding of AI and focuses on the common misconceptions about AI and the regulation of AI. The paper doesn't include any methodology or results, so I can't review it regarding these aspects.
The paper goes into a lot of detail with a very interesting take on AI regulation. Good use of referencing is used. No real concerns, except a few minor suggestions I have listed below.
1) The abstract mentions section 7 (line 17). But sections 5 and 6 are not mentioned. Consider including a sentence to mention what sections 5 and 6 are in the abstract.
2) There are no figures. This is a missed opportunity as they can help explain concepts. For example, in section 3, you discuss various cases/branches. First, please be more consistent in how you describe the alternative ideas you are discussing. Also, consider having a flow chart or figure/table that summarises these ideas to make it easier for the reader to understand and see the alternative ideas.
3) Another model is mentioned in section 5 (line 293). Consider including a figure that illustrates this somehow. Also, you mention the table of types of percepts by Norvig (line 286). Why not include the table and reference it?
In general, the paper is written well and covers interesting ideas/topics.
Comments on the Quality of English Language
The quality of English is very good. Only minor edits are needed. Just double-check grammar. No real concerns.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsI am honored to have reviewed this paper. I am rather pessimistic on how many people will able to understand it, as the level of illiteracy in the field of AI is growing faster than the technical capabilities of the AI systems themselves.
It is a great luck that the Editors will have this paper in their Special Issue.
There are many ideas and concepts that appear in the paper and which might be all worth to trigger discussions and a dialogue in the field.
The title ('AI and Regulations') might be good to be reconsidered by the author to allow a bigger audience of potential readers to access the paper.
Author Response
I thank you for your very generous reading of my paper. I agree that it would be useful to attract more reader but I must unfortunately admit that I have not been able to come up with a more attractive title.
Reviewer 4 Report
Comments and Suggestions for AuthorsThis is a very inspiring essay. The author compares artificial AI agents, alternative cognitive systems, with human actors, and characterizes the former as inhabiting a model of the world given by human actors. As such AI agents are limited in capabilities and the real dangers of AI agents are not that they will taking over the world but that the uneven power relationships between those who put them into use and those who are responsible to interact with them. Because the latter, however, are not allowed to employ discretion when interacting with AI agents.
As a reviewer I have three questions, a comment, and a suggestion to the author.
Q1: Seems that the author would agree that issues of privacy are real dangers and are proper to the use of AI systems (lines 161-2). If so, how would the author conceive the issue in his framework?
Q2: The author differentiates regulation from ethics by claiming that “Regulations seek to avoid certain consequences and to obtain (or at least to encourage) certain outcome” (lines 186-8). Defined as such, does not regulation amount to what consequentialism, an important faction of ethics, tries to achieve?
Q3: Does AI systems that are not explainable add anything new to the characterization of AI in sections 5 and 6?
Comment: I am wondering whether in this essay, the author hints a pessimistic view of the future for regulations of AI, for many who propose a moratorium on AI are business tycoons who are at the same time eager to use AI agents in their products and do not really care about whether those who develop and maintain AI agents have enough discretion power.
Suggestion: An elaboration of sections 5 and 6 of their link to the whole essay may make it clearer about their contribution to the main argument.
Comments on the Quality of English LanguageA typo GTP(line 323) and lapses(lines 16-7) need editing.
Author Response
Please view attached file.
Author Response File: Author Response.pdf
