Data-Driven Framework for Aligning Artificial Intelligence with Inclusive Development in the Global South

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper focuses on the core pain points of the disconnect between artificial intelligence development and inclusive goals in the global South, and innovatively proposes a four-dimensional data-driven framework with accessibility, subjectivity, accountability, and adaptability as the core. The study adopts a mixed method design, integrating 1920 large sample surveys, qualitative interviews, and participatory seminar data from 6 urban and rural sites, ultimately forming a complete research chain of "theoretical construction empirical testing mechanism explanation policy transformation".

The modification suggestions are as follows:

(1) There is ambiguity in the definition of core concepts in the paper: ① The operational definition of Agency does not distinguish between individual and community levels, and it is doubtful whether measurement indicators (such as "use of appeal channels") fully cover both levels; ② Adaptation does not specify the specific dimensions of local adaptation (such as language adaptation alone or cultural customs and infrastructure adaptation); ③ The types of "Harms" are not defined (such as exclusion caused by algorithm bias, privacy risks caused by data breaches), and the association between different types of harm and the four-dimensional framework is not distinguished. Suggestion: The author should supplement the corresponding table of "Concept Operational Definition Measurement Indicators" to clarify the attribution level of each indicator; Refine the local adaptation dimension of 'Adaptation'; ③ Classify and analyze the correlation strength between different types of injuries and framework dimensions to avoid measurement bias caused by conceptual ambiguity.

(2) The framework defines adaptation as the system's adaptation to local languages and practices, based on the cultural and infrastructure diversity of the global South. However, from the perspective of cross-cultural design theory, adaptation also needs to cover the adaptation of local power structures. How to demonstrate the coverage of the current Adaptation definition for non-technical dimension adaptation, it is necessary to explain the rationality of the definition boundary and avoid practical deviations caused by narrow adaptation dimensions.

(3) The structural equation model shows that the standardization coefficient between Access and service coverage is 0.41, and the use of a gap may weaken the actual effectiveness of Access. Why was the use of gap adjustment variables not introduced.

(4) The accountability dimension focuses on transparency and accountability tracing, based on the principle of accountability in AI ethics. However, in some regions, there is a lack of institutional capacity, such as a shortage of AI audit personnel. How to demonstrate the adaptability of current accountability indicators to areas with weak governance capacity, it is necessary to compare the difference in accountability scores between high governance capacity sites and low governance capacity sites, explain indicator adjustment strategies, and avoid dimension failure caused by neglecting governance capacity.

(5) In order to make introduction more comprehensive, it is recommended that the author refer to the following papers on algorithms. 1)Towards visual interaction: hand segmentation by combining 3d graph deep learning and laser point cloud for intelligent rehabilitation. 2)Intelligent rehabilitation in an aging population: empowering human-machine interaction for hand function rehabilitation through 3d deep learning and point cloud.

(6) Exploratory factor analysis supports a four factor structure, but from a measurement theory perspective, different cultural backgrounds in the global South may lead to variations in factor structure (such as differences in understanding of Agency between Africa and Asia). How to verify the cross-cultural invariance of factor structures. Multiple sets of confirmatory factor analyses are needed to compare the differences in factor loadings among different cultural sites, demonstrate the necessity of structural stability for cross regional framework reuse, and avoid insufficient theoretical generalization due to cultural differences.

Author Response

Dear Madam/Sir

Add Added in the word file.

Thank you!

Arjun

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

< !--StartFragment -->

Your paper makes a valuable and timely contribution by introducing and empirically validating the 4A framework, which is both innovative and relevant. The study is methodologically sound and clearly written. I recommend minor revisions to strengthen the alignment with Sustainability Journal’s scope, particularly by highlighting environmental sustainability linkages, condensing the literature review for clarity, and further emphasizing the unique contribution of the framework beyond existing models:-

1. Your literature review is indeed very comprehensive. You have based your framework on published works on topics such as inclusivity, inequality and governance.
2. While inclusivity, governance, and data justice are well covered, the manuscript could more explicitly connect its findings to environmental sustainability and SDG agenda. This would enhance fit with Sustainability.
3. The review is comprehensive but at times lengthy. Streamlining some sections would improve clarity, highlight the key scholarly gaps, and sharpen the contribution of the paper. You are encouraged to use tabular structures to reduce the lengthy paragraphs.
4. Your proposed framework indeed has a novelty factor, but the manuscript would benefit from a stronger statement of how the 4A model differs from or improves upon existing ICT4D or AI governance frameworks, ensuring its added value is transparent to readers.
5. The policy implications are fairly covered, but could be further tied to sustainability objectives, for example by suggesting green ICT procurement standards or accountability mechanisms that also address environmental risks.
6. I noticed that there were some inconsistencies in the referencing. It appears that there was some redundancies or duplication in the reference list. This cannot be professionally acceptable, hence kindly try to fix this issue.

The suggested revisions are modest and do not question the research quality. However, if you address them, it will enhance the clarity and scope of your work, allowing the manuscript to make a strong contribution to Sustainability and serve as a valuable resource for both academic and policy audiences focused on AI, inclusivity, and sustainable development.

< !--EndFragment -->

Author Response

Dear Madam/Sir

Add Added in the word file.

Thank you!

Arjun

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The author answered the questions raised by the reviewer very well, and this paper is acceptable.