Chances and Risks of Artificial Intelligence—A Concept of Developing and Exploiting Machine Intelligence for Future Societies
Round 1
Reviewer 1 Report
This paper tries to deliver the risks of AI in future societies but lacks in several aspects.
The abstract is huge and looks like a chapter, whereas it should be about 150 words summarising what the paper is about.
In the paper there aren't proper chapters/sections like introduction, background but there are other introductory sections that do not make much sense. The validation section does not validate anything really.
The paper is full of bullet points and not justified decisions while some figures are unreadable or do not make any sense.
Author Response
Chances and Risks of Artificial Intelligence-A concept of developing and exploiting machine intelligence for future societies-
Reinhard Hutter 1and Marcus Hutter 2,*
Review Round 1,
received 21 April 2021
responded 27 Apr. 2021
Response to Reviewer 1
Comments
Point 1: The abstract is huge...
Response 1: The abstract has been substantially reduced in size. Parts of the deleted wording will be used for extending the Introduction chpt. as requested by Reviewer #2.
Point2: ....no proper chapters like inroduction, background, limited validation
Response 2: The introduction (chpt.1) has been substantially improved and amended by referring to the individual chapters. Background of the AI discipline is given in chapter 2. As to Validation (Chapter4), this is not comparable to software or other technical validations. As we are talking here about future concepts not yet technically realized, validation here is of qualitative nature on human vs. artificial capabilities, the possible impact of new CI solutions in economy and politics, and on data requirements.
Point 3:...full of bullet points
Response 3: I think, bullet points are well suited to structure complex subjects. Formulating all in floating text would substantially reduce oversight, transparency and would unnecessarily increase the size of the article.
Point4: ...full of...not justified decisions
Response 4:: The paper does not really talk about decisions. All we discuss and recommend are ideas and concepts for the future
Point5: Minor spell-checks
Response 5: Spell checks have been done throughout the document, using UK version of Microsoft spell checker
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors presented a survey about the different risks of application of artificial intelligence. The paper is well written and an in deep review about the recent advancements on the topic. I have some minor comments.
- Introduction is very short, one paragraph
- the authors need to provide a very good Introduction to introduce their work before going through the details in the next sections. They need to review the different limitations by analyzing the existing survey papers and highlighting the missing reviews in order to include them, also, present the motivations behind this work and why it can be considered for publication (novelty)
- figures and their included text are too big compared to the text in the paper
- literature review from 1980 is not needed as e are in 2021 and we need to focus more on recent works and recent advancements by discussing research in AI in the past three years
- for nano and bio technologies, how the AI can be operated to replace the human intervention ? may be an example can be provided in healthcare/remote surgery
- despite the advantages of the AI, it is known that it is will open the door to severs security problems and concerns, i would like to see a section about how security can be achieved while exploiting all this intelligence in our lives (data security, communications, privacy)
- references need to be imperatively updated with recent ones instead of papers from 15 years ago (2005/2006 as example)
Author Response
Chances and Risks of Artificial Intelligence-A concept of developing and exploiting machine intelligence for future societies-
Reinhard Hutter 1and Marcus Hutter 2,*
Review Round 1,
received 21 April 2021
responded 27 Apr. 2021
Response to Reviewer 2
Comments
Point 1: Provide a very good Introduction, ...review the different limitaions..., ...present the motivations... (novelty).
Response 2: The introduction (chpt.2) has been substantially extended accordingly, including the novelty character of the subject as analyzed here, the motivation for publication and the framing limitations. At the end of the introduction short reference is given to the individual chapters of the paper.
Point 2: Figures are too big compared to the text...
Response 2: Figures have been reduced and it is up to the editor to choose the proper size. Figures are considered self-explanatory to the accompanying text. Less text in the figures therefore would require unnecessarily more floating text.
Point 3: literature review from 1980 is not needed, ...discussing research in the past three years.
Response 3: The discussions going back to 1980 should demonstrate that (a) contrary to early predictions, it was quite a long process until AI developed to the capabilities as of today, and (b) the authors have a broad and in-depth background of the subject. A view of the "last 3 years" on the subject would too much limit the scope. Some specific methodologies/qualities in AI such as self learning, neural networks, complex matematical modelling require many years from idea to solution. This longer-term view appears to be neccessary in order to understand that AI – compared to its earlier stages -is now at the edge of substantial breakthrough for applications in the various domains of society, politics and economy. See some changes in chapter 2.
Point 4: for nano and bio technologies, how the AI can be operated to replace the human intervention ?, ...an example...
Response 4: Five examples from Biology/surgery and Nano-technologies are given in Table 1. Some amendments are now introduced in Table 1, also indicating ease of human intervention. Bio & Nano are not the target disciplines of this special ASI issue. If you want to expand on these, we would create a further chapter which, however, might be somehow out of scope of the ISR issue.
Point 5: ... a section about how security can be achieved while exploiting all this intelligence in our lives...
Response 5: There is chapter 6.2 on risk management and security. The list of bullets expressing the benefits of AI/CI in security have been amended to the comment of the reviewer, indicating application orientation.
Point 6: Refernces need to be ...updated
Response 6: The majority of references is from the last decade, up to 2020. The aspect of ethical responsibility may be underrepresented in the article. I added an extensive reference on Ethics in Artificial intelligence that includes robot ethics, AI/ Cyber threats to society, AI accountability, danger of bias, AI liability, military use, machine-ethics. As to older references, please see my comment to Point3 above.
I would particularly draw attention to the important present-day activities as referenced by [11], [19] and [28].
Author Response File: Author Response.pdf
Reviewer 3 Report
The paper is interesting, however I would suggest the author an important aspect/limitation of AI, that is the possibility to create (new) ideas.
Something (creativity) about we humans have the possibility to do, when new bright ideas can come when looking at the children, at the sea, at the peak of the mountains...or even cooking.
In my opinion , I would not stress that "Working 40 hours week or so is not a law of nature!", I would rather say that creative works will be more needed, while some works will simply disappear (as happens already..) being replaced by automatic and/or AI-enabled robots. So, people will be more and more occupied in works requiring both technical issues (and here AI is helpful with no doubt) but also creating new things from scratch. Such kind of works will require 40 hours week, or even more (depends on the area and on the personal interests).
Author Response
Chances and Risks of Artificial Intelligence-A concept of developing and exploiting machine intelligence for future societies-
Reinhard Hutter 1and Marcus Hutter 2,*
Review Round 1,
received 21 April 2021
responded 27 Apr. 2021
Response to Reviewer 3
Comments
Point 1: Something (creativity) about we humans have the possibility to do, when new bright ideas can come when looking at the children, at the sea, at the peak of the mountains...or even cooking.
Response 1: Discussing "Creativity", feelings, self-consciousness etc. of AI is important and challrnging, and is an ongoing research area with still many unsolved or uncertain areas. It is an interdisciplinary field involving computer/AI science, sociology, philosophy, neurology/ brain research, psychology and more. We have limited ourselves here to the benefits of AI for societies with some focus on risks and security. Creative (however more technical options) in security are discussed in chapter 6.2, some remarks on creativity, intuition, arts etc., please see also chpt. 4.1 and in Tabl 1 under the domain Education.
Point 2: ...I would not stress that "Working 40 hours week or so is not a law of nature!"
Response 2: Yes, this is very true! I have modified the wording in chpt. 4.3 accordingly.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
No further comments
Author Response
Thanks for your time and valuable comments
