The SP Theory of Intelligence, and Its Realisation in the SP Computer Model, as a Foundation for the Development of Artificial General Intelligence
Round 1
Reviewer 1 Report
Seems only sp-ratio is used to compare SPTI with ALTs. Is there any other criteria?
Author Response
I should explain that the whole article has been radically revised. But at the same time I have taken account of comments by reviewers and give my responses to Reviewer 1 below.
Many thanks for this positive review.
Regarding the comment "English language and style are fine/minor spell check required", I have given the whole revised document a thorough spell check with the WinEdt spell checker and have made corrections where necessary.
Regarding this comment: "Seems only sp-ratio is used to compare SPTI with ALTs. Is there any other criteria?", the revised version of the MS uses four main criteria for evaluating the SPTI and the ALTs:
* The Simplicity of the system.
* The Power of the system in expressing aspects of human-like intelligence.
* Any other strengths or weaknesses of a given system as a foundation for the development of AGI.
* A sum of the preceding three ratings.
This is described in Section 3.1 of the revised submission.
Reviewer 2 Report
The author proposes Foundation of Development of AGI as alternative of SPTI. A review of existing systems for AGI is made. Different systems are evaluated in terms of several criteria. The paper is very interesting and is written in understandable way. It can be useful for scientist from different fields. I suggest the acceptance of this paper.
Author Response
I should explain that the whole article has been radically revised. But at the same time I have taken account of comments by reviewers and give my responses to Reviewer 2 below.
Many thanks for this positive review.
Regarding the comment "English language and style are fine/minor spell check required", I have given the whole revised document a thorough spell check with the WinEdt spell checker and have made corrections where necessary.
Reviewer 3 Report
Title: The SP Theory of Intelligence, and Its Realisation in the SP Computer Model, as a Foundation for the Development of Human-Level Broad AI, Aka Artificial General Intelligence
Comments: The article is well written and presented. The quality of the statements is up to the mark as per journal standards. However, some minor suggestions may be incorporated for further increasing the manuscript quality.
a. The author may include theoretical comparison (preffereably tabular) and discuss the potentials.
b. Discuss more on "Old SP-patterns" and "New SP-patterns".
c. Some of the sentences needs in-depth discussions.
1. The SP Machine will be hosted on a high-end workstation with one or more 206 GPUs, or equivalent facilities in the cloud.
2. Obviously, each system should have strengths that are relevant to the development of human-like intelligence
3. Similar things can be said about Soar and ACT-R. Each of them is a generalised framework that must be ‘programmed’ to create any one of a variety of specific systems.
4. Hence, in this narrow evaluation, the SPTI has a small s, meaning great Simplicity.
Author Response
I should explain that the whole article has been radically revised. But at the same time I have taken account of comments by reviewers and give my responses to Reviewer 3 below.
Regarding the comment "English language and style are fine/minor spell check required", I have given the whole revised document a thorough spell check with the WinEdt spell checker and have made corrections where necessary.
Other points:
* "The author may include theoretical comparison (preferably tabular) and discuss the potentials." These points should now be covered by the thorough revision of the article mentioned above.
* "Discuss more on 'Old SP-patterns' and 'New SP-patterns' ". I've tried to add relevant information in Appendix A to clarify the roles of New SP-patterns and Old SP-patterns.
* "Some of the sentences needs in-depth discussions." Which sentences?
* "The SP Machine will be hosted on a high-end workstation with one or more 206 GPUs, or equivalent facilities in the cloud." Yes, that it what is envisaged for future development of the SPCM.
* "Obviously, each system should have strengths that are relevant to the development of human-like intelligence." Yes, this is emphasised in the revised submission, although strengths in non-AI areas are considered to add to the plausibility of the SPTI as a foundation for the development of AGI.
* "Similar things can be said about Soar and ACT-R. Each of them is a generalised framework that must be ‘programmed’ to create any one of a variety of specific systems." Yes.
* "Hence, in this narrow evaluation, the SPTI has a small s, meaning great Simplicity." Yes, although the evaluation metric in the revised submission is different from that in the original submission.
