From Static Prediction to Mindful Machines: A Paradigm Shift in Distributed AI Systems

Round 1

Reviewer 1 Report (Previous Reviewer 3)

Comments and Suggestions for Authors

The article introduces a detailed, multilayered conceptual framework built on the unique properties described in Burgin’s General Theory of Information, the Burgin – Mikkilineni Thesis, Deutsch’s Epistemic Thesis, and Hill’s Fold Theory. This framework makes it possible to construct interacting agents from a small, structurally declarative core — the Digital Genome — which stores all information required to create and evolve these agents, defines the boundaries of their potential development and interaction, and is is supported an operating system AMOS (Autopoietic and Meta-Cognitive Operating System) designed to work with this core and manage the resulting multi-agent system.

This conceptual foundation is strong enough for publication in Computers magazine. However, the illustrative example used in the article — the classic Credit Default Prediction problem—does not demonstrate the strengths of the proposed framework, because it does not take advantage of its unique capabilities. To showcase these features, the example should include an additional method for solving the prediction task alongside logistic regression, along with a scenario in which the system switches from one method to the other in response to changing conditions.

Author Response

Please see attachment

Author Response File: Author Response.pdf

Reviewer 2 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

Positive points

• The paper introduces an innovative architectural model — the Mindful Machine — that advances beyond traditional AI paradigms by embedding self-regulation, memory, and governance directly into computation.
• The integration of the General Theory of Information, Burgin–Mikkilineni Thesis, Deutsch’s Epistemic Thesis, and Fold Theory gives the work a rigorous philosophical and scientific grounding.
• The implementation of the credit-default prediction system using the AMOS platform effectively bridges theory and practice, demonstrating feasibility in a real-world domain.
• The emphasis on transparency, traceability, and coherence directly addresses major limitations of current black-box AI systems, contributing to the growing field of trustworthy AI.
• The manuscript is well-organized — from theoretical foundations to implementation and comparative analysis — making it accessible to both technical and conceptual readers.

Points which may need improvement

• The paper describes the architecture well. The implementation aspects need detail. The implementation aspects could include comparison metrics, scalability benchmarks or empirical validation, beyond the case study. Adding detail to the implementation aspects would make the scientific rigor stronger.
• Language and Clarity: The manuscript’s English is generally solid, but certain sections (especially theoretical explanations) are dense. Simplifying complex sentences and improving flow would enhance readability for a broader audience.

 

Author Response

Please see attachment

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

I am disappointed to say that this is a submission that I find easy to reject. This is a pity, as I think there are some potentially very interesting ideas here. The main issue is that it is long on description but short on evidence. Using a specific problem like the credit-default prediction problem to analysis differences is a good idea; unfortunately this is not really done. So what remains is a lots of interesting ideas and strong claims without any real evidence, even in the concrete scenario discussed, to back them up. 

A walk through the paper is as follows. 

Section 2 briefly describes 4 key concepts, being General Theory of Information (GTI), the Burgin–Mikkilineni Thesis (BMT), Deutsch’s epistemic framework of knowledge, and Fold Theory. While these all appear interesting, they are hardly universally known and understood, and hence would benefit from a much longer and deeper introduction that is given here. As it is, it seems there is enough here to whet the appetite, but nothing more. 

Section 3 is an architectural description of the implemetantion, and Section 4 is a brief commentary on it. Section 5 identifies some limitations of previous solutions to the credit-default prediction problem and states in a high level manner how this approach overcomes them. However there is little or no analysis or technical discussion here, being basically Table 1 which lists a variety of properties without any discussion of how these are achieved. So between them, Sections 3, 4 and 5 describe a solution that is close to incomprehensible, and make claims about it without any real evidence.  

Section 6 provides conclusions and further work, restating various previous claims. There is also an Appendix with a description of the implemention of this procedure together with a workflow diagram. But that does not analyse or discuss how this improves on previous approaches to this problem. 

So the impression I am left with is that there are plenty of good ideas here, seemingly, but little description of how these work in practice, and no insight into how the application of these makes for a better solution than existing ones. 

A more minor comment is the following one. "Post-Turing computation" seems to be a rather meaningless phrase and should not be used. Would one ever use the dual phrase of "pre-Turing computation"? I think one could also take issue with the statement that a key property of Turing machines is "the separation of the computer from the computed". One could argue that Universal Turing machines assume exactly the opposite. So in short I think it is better to drop any implicit comparison with Turing machines and focus on the novel properties of this approach. 

Author Response

Section 2 briefly describes 4 key concepts, being General Theory of Information (GTI), the Burgin–Mikkilineni Thesis (BMT), Deutsch’s epistemic framework of knowledge, and Fold Theory. While these all appear interesting, they are hardly universally known and understood, and hence would benefit from a much longer and deeper introduction that is given here. As it is, it seems there is enough here to whet the appetite, but nothing more. 

Completely rewritten with Expanded explanation and references

Section 3 is an architectural description of the implementation, and Section 4 is a brief commentary on it. Section 5 identifies some limitations of previous solutions to the credit-default prediction problem and states in a high level manner how this approach overcomes them. However, there is little or no analysis or technical discussion here, being basically Table 1 which lists a variety of properties without any discussion of how these are achieved. So between them, Sections 3, 4 and 5 describe a solution that is close to incomprehensible and make claims about it without any real evidence.  

Rewritten to include explanations screenshots of components and results

Section 6 provides conclusions and further work, restating various previous claims. There is also an Appendix with a description of the implemention of this procedure together with a workflow diagram. But that does not analyse or discuss how this improves on previous approaches to this problem

Rewritten with a comparison of results between our approach and traditional Machine Learning with Logistic Regression using same data was used in the textbook. Figures and discussion provide our insights from this exercise

A more minor comment is the following one. "Post-Turing computation" seems to be a rather meaningless phrase and should not be used. Would one ever use the dual phrase of "pre-Turing computation"? I think one could also take issue with the statement that a key property of Turing machines is "the separation of the computer from the computed". One could argue that Universal Turing machines assume exactly the opposite. So in short I think it is better to drop any implicit comparison with Turing machines and focus on the novel properties of this approach.

We are referring Cockshott et al in the book Cockshott, P.; MacKenzie, L.M.; Michaelson, G. Computation and Its Limits; Oxford University Press: Oxford, UK, 2012 p. 215 "The concept of the universal Turing machine has allowed us to create general-purpose computers and [19] (p. 215) “use them to deterministically model any physical system, of which they are not themselves a part to an arbitrary degree of accuracy. Their logical limits arise when we try to get them to model a part of the world that includes themselves.” See also Mikkilineni, R. (2022). Infusing Autopoietic and Cognitive Behaviors into Digital Automata to Improve Their Sentience, Resilience, and Intelligence. Big Data and Cognitive Computing, 6(1), 7. https://doi.org/10.3390/bdcc601000. Turing computing model requires a third party to manage computing structure to deploy the application structure (application components and their connections). An autopoietic system models both computing structure (IaaS and PaaS) and the application structure and manages with APM, CNN, SWM and SEM Operating system components. Thus the model contains the computer and the computed. Fold Theory says that ontological structures (the observed) and epistemic structures (the observer) interact with each other to form coherent structure.

 

Reviewer 2 Report

Comments and Suggestions for Authors

The paper is a new and well-theorized concept that effectively brings together self-regulation, meta-cognition, and distributed system design. Strengthening the methodology section with better specified validation metrics and tightening up the language for conciseness would give it greater impact. It is a good conceptual contribution that deserves to be published but references can be improved by providing links for each.

Author Response

The paper is a new and well-theorized concept that effectively brings together self-regulation, meta-cognition, and distributed system design. Strengthening the methodology section with better specified validation metrics and tightening up the language for conciseness would give it greater impact. It is a good conceptual contribution that deserves to be published but references can be improved by providing links for each.

Rewrote methodology and gave comparison with traditional ML with logistic regression with same data. References are reexamined and links checked for accuracy.

Reviewer 3 Report

Comments and Suggestions for Authors

1. The authors propose a new conceptual framework for the description and operation of distributed systems, which they present as a post-Turing computational paradigm. Systems designed under this framework are referred to as Mindful Machines and are intended to operate within a specialized AMOS operating system.

2. The high-level descriptions of both the framework and the operating system are clear, and the claimed capabilities appear promising.

3. However, the presentation lacks sufficient detail, as the components are treated as black boxes with only declared sets of properties. The provided default prediction case study, unfortunately, does not help to clarify these internal mechanisms.

4. For AMOS, it would be helpful to see the internal structure of its layers and how these layers interact.

5. Likewise, the Mindful Machine concept would benefit from a more detailed depiction of its architecture and its integration within AMOS.

6. The diagrams in Figures 1–5 do not provide additional clarity, as they themselves require further explanation.

7. As for minor issues, the caption “Figure 4: Workflow Diagram for GTI-based Credit Default Prediction Implementation” is repeated on page 8.

8. In conclusion, the article requires further refinement — particularly in elaborating the proposed concepts, which currently appear overly general in contrast to the Turing computational paradigm, whose structure is formal and well defined.

 

Author Response

. However, the presentation lacks sufficient detail, as the components are treated as black boxes with only declared sets of properties. The provided default prediction case study, unfortunately, does not help to clarify these internal mechanisms.

Rewritten all sections to provide more detail and a comparison with traditional ML with Logistic Regression is provided with same data.

. For AMOS, it would be helpful to see the internal structure of its layers and how these layers interact.

5. Likewise, the Mindful Machine concept would benefit from a more detailed depiction of its architecture and its integration within AMOS.
6. The diagrams in Figures 1–5 do not provide additional clarity, as they themselves require further explanation.

Sections are rewritten to provide more details of schema and the instances. The explanations are expanded.

7. As for minor issues, the caption “Figure 4: Workflow Diagram for GTI-based Credit Default Prediction Implementation” is repeated on page 8.
8. In conclusion, the article requires further refinement — particularly in elaborating the proposed concepts, which currently appear overly general in contrast to the Turing computational paradigm, whose structure is formal and well defined.

Corrected the repetition. Provided more insights learned from this exercise.

 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I can see that some effort has gone into the presentation of the paper, with more explanations and details at key points. But the fundamental issue remains, which is that this is necessarily a descriptive summary of four very interesting but complex theoretical developments realised in an implementation. This format is simply too short to do justice to the ideas involved. Were this a case study focused on the specific credit default prediction problem, identifying weaknesses of current approaches and how they may be overcome, that would be more interesting. As it is, it leaves me feelling that I need to read several other longer papers to even begin to understand how this is an improvement on current practice. 

Author Response

Author’s Response to Reviewer 1

We appreciate your engagement with our manuscript and the effort you’ve made to assess its contributions. We recognize that the ideas presented challenge conventional formats and expectations, and we welcome the opportunity to clarify the intent and scope of our work.

This paper does not aim to incrementally improve existing AI techniques or present a narrowly scoped case study. Instead, it introduces a new class of distributed software systems — Mindful Machines — that embody a self-regulating, knowledge-centric architecture. The goal is to transcend legacy computational paradigms and offer a scalable framework for ethically aligned, epistemically transparent AI.

We understand that paradigm shifts often provoke discomfort, especially when they require new scaffolds of understanding. The need to consult additional works is not a limitation of this paper, but a reflection of its role as a gateway to a broader epistemic transition. We invite reviewers to engage with the work not solely through the lens of current practice, but as a contribution to the evolution of foundational knowledge.

With this framing in mind, we offer the following point-by-point responses to your comments:

Reviewer Comment:

“This is necessarily a descriptive summary of four very interesting but complex theoretical developments realised in an implementation. This format is simply too short to do justice to the ideas involved.”

Author Response: We agree that the ideas presented are complex and foundational. However, the intent of this paper is not to exhaustively explain each theoretical development, but to formally introduce a new class of distributed software systems — Mindful Machines — that unify computation and cognition within a self-regulating architecture. This is not a descriptive summary, but a ratification of a paradigm shift. We acknowledge that deeper engagement may require reference to prior works, and we view this paper as a gateway rather than a standalone explainer.

Reviewer Comment:

“Were this a case study focused on the specific credit default prediction problem… that would be more interesting.”

Author Response: We respectfully suggest that this framing misses the purpose of the paper. The credit default prediction example is not the focal point, but a comparative lens to highlight the limitations of legacy AI approaches and the advantages of knowledge-centric architectures. Our goal is not to improve existing heuristics, but to demonstrate a new epistemic scaffold for distributed intelligence. A case study would constrain the scope and dilute the paradigm-level implications.

Reviewer Comment:

“It leaves me feeling that I need to read several other longer papers to even begin to understand how this is an improvement on current practice.”

Author Response: We appreciate this honest reflection. The need to consult additional materials is not a flaw, but a symptom of epistemic transition. Paradigm shifts often require scaffolding across multiple works. We have made efforts to include references and contextual bridges, and we welcome future opportunities to expand this framework in more pedagogical formats.

Additional Clarification: This paper is not an incremental improvement of current technology. It challenges the foundational assumptions of classical computer science and proposes a new path forward — one that is ethically aligned, epistemically transparent, and architecturally self-regulating. We invite reviewers to engage with the work not as a technical enhancement, but as a philosophical and architectural reorientation.

Closing Note: We recognize that the theoretical foundations referenced in this paper span multiple disciplines and may require additional context. For reviewers wishing to accelerate their understanding of these frameworks and their relationship to the Mindful Machine Architecture, we suggest leveraging large language models (LLMs). These models have access to the relevant literature and can assist in mapping the conceptual terrain, offering summaries, comparisons, and connections that may support deeper engagement with the paradigm we propose.