Frequently Asked Questions

HIKMA 2025 is a live experiment in AI-driven scholarship. It tests the role of AI across the research cycle: generating papers, conducting reviews, revising manuscripts, and presenting findings.

The conference examines whether AI can responsibly contribute to science while maintaining human oversight, ethical safeguards, and academic integrity.

HIKMA 2025 follows a structured methodology to demonstrate the role of Artificial Intelligence in the complete academic research cycle. The process consists of the following steps:

This methodology ensures that every stage of scholarly communication—generation, review, revision, and dissemination—was carried out by AI under structured human oversight.

All submissions are generated by AI systems, using prompts and themes defined by the organizers.

Human participants may design prompts, define research directions, and provide oversight. Core paper generation, however, must be done by AI.

No. HIKMA 2025 only accepts papers produced primarily by AI.

Papers must align with one of the five conference tracks: Artificial Intelligence, Precision Health, Progressive Education, Sustainability, or Social Progress.

Participants provide prompts or research directions. AI systems generate manuscripts, which are submitted through the conference platform.

Yes. After AI-based review, selected submissions undergo revision cycles managed through AI Scholar Frontier with human oversight.

Peer review is conducted by AI models trained on academic rubrics. Human experts oversee top-ranked submissions for final selection.

Yes. Reviews and metadata about the AI models used will be openly available.

Errors are expected and will be part of the learning process. Community feedback and human oversight will be used to identify and study them.

The AI paper generation follows a structured pipeline: 1) Human researchers provide research prompts and dataset specifications, 2) AI systems analyze relevant literature and generate research questions, 3) AI conducts experiments or analysis on the specified datasets, 4) AI writes the complete manuscript in LaTeX format with proper citations, 5) The paper undergoes automated quality checks before submission.

AI papers are generated using publicly available research datasets relevant to each track. These include medical imaging datasets for Precision Health papers, educational assessment data for Progressive Education research, environmental monitoring data for Sustainability studies, and social survey data for Social Progress research. All dataset usage follows ethical guidelines and proper attribution.

AI reviewers are trained on established academic review criteria including methodological rigor, novelty of contribution, clarity of presentation, and ethical considerations. Each paper receives multiple independent AI reviews using different model configurations. The review process evaluates technical soundness, reproducibility, significance of findings, and alignment with conference themes.

After receiving AI-generated reviews, the system automatically revises papers by: 1) Analyzing reviewer feedback and identifying specific improvement areas, 2) Modifying methodology, analysis, or presentation based on comments, 3) Adding missing references or clarifying unclear sections, 4) Generating a detailed response letter addressing each reviewer concern, 5) Creating a revised manuscript that tracks all changes for transparency.

Final acceptance decisions combine AI reviewer scores with human oversight. An AI conference chair reviews all submissions and reviewer comments, then makes recommendations. Human organizers provide final approval, focusing on ensuring ethical standards, methodological soundness, and alignment with conference goals. This hybrid approach maintains academic standards while testing AI capabilities.

Accepted papers are presented by AI avatars that generate presentation slides, deliver spoken narration, and respond to audience questions in real-time. Each avatar is customized to match the research domain and presents with appropriate academic style and terminology. Presentations include visual aids, data visualizations, and interactive elements to engage the audience.

Yes, AI avatars are equipped with deep knowledge of their research papers and can engage in Q&A sessions. They can explain methodology details, discuss limitations, compare with related work, and address technical questions. However, they are programmed to acknowledge when questions exceed their knowledge boundaries and defer to human moderators when appropriate.

AI avatar presentations offer consistent delivery quality, perfect timing, and the ability to access and cross-reference vast amounts of literature instantly. Unlike human presenters, AI avatars can provide immediate statistical calculations, generate visual comparisons on-demand, and present in multiple languages. They also maintain detailed logs of all interactions for post-conference analysis.

AI avatars are programmed with strict accuracy protocols: 1) They can only present information directly from their research papers, 2) All claims are linked to specific sections and references, 3) They explicitly state when making assumptions or extrapolations, 4) Human moderators monitor presentations for factual accuracy, 5) Post-presentation fact-checking validates all presented information.

The design ensures human oversight at all stages. The experiment tests augmentation of human scholarship, not replacement, with safeguards for transparency, accountability, and academic integrity.

No, because HIKMA 2025 operates with full transparency about AI involvement. All papers are clearly labeled as AI-generated, the process is documented, and human oversight is maintained throughout. This differs from undisclosed AI use, which would violate academic integrity. The experiment explicitly tests whether AI can contribute to scholarship under ethical frameworks.

AI bias is a recognized limitation that HIKMA 2025 actively studies rather than ignores. We: 1) Use diverse AI models to reduce single-source bias, 2) Require human validation of research findings, 3) Document all AI training data sources, 4) Include bias detection as part of the review process, 5) Publish bias analysis as part of post-conference findings. The goal is understanding AI limitations, not claiming AI perfection.

This is a key area of investigation for HIKMA 2025. We implement multiple safeguards: 1) AI systems are constrained to use only verified datasets, 2) All citations are automatically verified against academic databases, 3) Data analysis results are reproducible through provided code, 4) Human experts fact-check all accepted papers, 5) Any fabrication discoveries become part of the published experimental results. The purpose is learning about AI limitations, not avoiding them.

HIKMA 2025 is designed to complement, not replace, traditional research. It's clearly labeled as an experimental conference testing AI capabilities. Real researchers benefit by: 1) Learning about AI tools for research acceleration, 2) Understanding AI limitations before widespread adoption, 3) Participating in shaping ethical AI use policies, 4) Accessing insights about human-AI collaboration. This is research about AI in research, not AI replacing research.

HIKMA 2025 directly addresses this concern by providing empirical data about AI research capabilities and limitations. Rather than letting AI research develop without oversight, we're studying it systematically. The results will inform policy decisions, ethical guidelines, and best practices for human-AI collaboration in academia. The experiment helps prepare the research community for informed decision-making about AI integration.

Attribution in AI-generated research follows strict protocols: 1) AI systems are clearly credited as primary authors, 2) Human prompters and supervisors receive appropriate recognition, 3) All training data sources are acknowledged, 4) Original dataset creators are cited, 5) Human oversight contributors are listed, 6) The specific AI models and versions used are documented. This creates a new model for attribution that respects all contributors to the research process.

Safety protocols are embedded throughout the process: 1) Research topics are pre-screened for potential harm, 2) AI systems cannot access dangerous or restricted information, 3) Human experts review all findings before publication, 4) Ethics review boards evaluate all accepted papers, 5) Harmful content is flagged and either rejected or published with appropriate warnings and context. The goal is responsible exploration of AI capabilities within ethical boundaries.

Yes. The conference will publish a collective analysis of AI performance, reviewer reliability, and human–AI collaboration outcomes.

For questions, please contact the organizing committee at info@hikma2025.org.