Manuel B. Garcia

Manuel B. Garcia serves as the Senior Director for Educational Technology and Digital Learning at FEU Institute of Technology, Manila, Philippines. Read More

Contact Info

1607, FEU Tech Building,
P. Paredes St, Sampaloc,
Manila, Philippines
mbgarcia@feutech.edu.ph

Follow Me

The Complete Library

All Research Guides

Browse every published guide in the Researcher's Field Guide, including the complete research journey and supplemental collections, listed alphabetically for quick reference.

Complete Library

Every published guide

Showing 351–400 of 5,065 guides

5,065 guides
PART 0D · SUBPART 62 · GUIDE 846

AI and Scientific Image Falsification

Can Generative AI Facilitate Scientific Image Falsification?

Generative AI can create or alter scientific images in ways that misrepresent experimental evidence. Learn when AI-assisted image processing becomes falsification, how legitimate enhancement differs from manipulation, and what researchers should document to protect image integrity.

PART 0D · SUBPART 62 · GUIDE 845

Generative AI and Research Data Fabrication

Can Generative AI Facilitate Research Data Fabrication?

Generative AI can produce realistic-looking research data, making fabrication easier to carry out and potentially harder to recognize from the final dataset alone. Understand how fabrication differs from legitimate synthetic data generation and what researchers should do to preserve the integrity of empirical evidence.

PART 0D · SUBPART 62 · GUIDE 844

Undisclosed AI Use and Research Misconduct

Is Undisclosed AI Use Considered Research Misconduct?

Failing to disclose AI assistance may violate journal or institutional policies, but it does not automatically constitute formal research misconduct. Understand when nondisclosure becomes misleading, how misconduct definitions apply, and what researchers should do when AI use was not reported.

PART 0D · SUBPART 62 · GUIDE 843

When AI Use Threatens Research Integrity

When Does AI Use Become a Research-Integrity Problem?

AI use becomes a research-integrity concern when it compromises the accuracy, honesty, transparency, or accountability of research. Learn how to distinguish acceptable assistance, questionable practices, policy violations, and formal research misconduct.

PART 0D · SUBPART 61 · GUIDE 840

Institutional AI Restrictions for Research

Can a Research Institution Restrict AI Use for Commercially Sensitive Research?

Research institutions may restrict the use of generative AI when research involves confidential inventions, proprietary information, or contractual obligations. Learn what institutions can regulate, why restrictions may apply, and how researchers can determine whether a particular AI workflow is authorized.

PART 0D · SUBPART 61 · GUIDE 839

AI-Assisted Code and Design Ownership

Who Owns AI-Assisted Research Code or Designs?

Using AI to write research code or create designs does not automatically determine who owns them. Copyright protection, human contributions, institutional agreements, AI service terms, and third-party licenses may all affect the rights researchers can claim.

PART 0D · SUBPART 61 · GUIDE 838

Can AI Be a Patent Inventor?

Can an AI System Be Named as an Inventor on a Patent?

AI systems cannot currently be named as patent inventors in major jurisdictions such as the United States and Europe. Learn why the DABUS cases matter, how human inventorship is assessed, and what researchers should document when AI contributes to an invention.

PART 0D · SUBPART 61 · GUIDE 837

Can AI Generate Inventive Ideas?

Can Generative AI Generate an Inventive Research Idea?

Generative AI can propose research ideas and technical solutions that may be novel or useful, but generating an unexpected suggestion is not the same as establishing an invention. Learn how to assess AI-generated ideas, distinguish technical creativity from legal inventorship, and determine what human researchers must contribute.

PART 0D · SUBPART 61 · GUIDE 836

Patentable Ideas and Public AI Systems

Should Researchers Enter Potentially Patentable Ideas Into Public AI Systems?

Researchers should generally avoid entering undisclosed, potentially patentable technical details into public AI systems without appropriate safeguards and authorization. Learn when AI assistance may be acceptable, what risks to assess, and how to obtain useful assistance without unnecessarily exposing an invention.

PART 0D · SUBPART 61 · GUIDE 835

Generative AI and Patentability

Can Using Generative AI Affect the Patentability of a Research Invention?

Generative AI assistance does not automatically prevent a research invention from being patented. However, human inventorship, novelty, inventive step, technical eligibility, and disclosure requirements remain important, and AI use can introduce complications that researchers should assess before filing.

PART 0D · SUBPART 61 · GUIDE 834

AI and Unpublished Invention Confidentiality

Can Entering an Unpublished Research Invention Into AI Harm Its Confidentiality?

Entering an unpublished research invention into an AI system may create confidentiality risks, but it does not automatically constitute public disclosure or destroy patent rights. Understand how AI service terms, data handling, and institutional obligations affect the protection of potentially patentable research.

PART 0D · SUBPART 61 · GUIDE 833

AI and Research Invention Ownership

Can Using AI Affect Ownership of a Research Invention?

Using AI in research does not automatically change who owns an invention, but inventorship, employment agreements, institutional policies, and contractual rights may affect the outcome. Learn how to distinguish these issues and protect potentially valuable research discoveries.

PART 0D · SUBPART 60 · GUIDE 832

Copyright Liability in AI-Assisted Research

Who Is Responsible for Copyright Infringement Caused by AI-Assisted Research Content?

Copyright infringement involving AI-assisted research may create responsibilities for researchers, institutions, publishers, AI providers, or other parties depending on their conduct and applicable law. Researchers should not assume that AI providers automatically bear liability for generated content. Clear permissions, source verification, institutional governance, and appropriate publication practices can help manage these risks.

PART 0D · SUBPART 60 · GUIDE 831

Publisher Restrictions on AI Processing

Can a Publisher Restrict AI Processing of Its Published Content?

Publishers may restrict certain forms of AI processing through copyright rights, licensing agreements, subscription contracts, and platform terms. However, their authority is not unlimited, and applicable statutory exceptions or open licenses may permit particular activities. Researchers should distinguish binding contractual restrictions from publisher guidance and verify the terms governing their actual access.

PART 0D · SUBPART 60 · GUIDE 830

Lawful Research Paper Access and AI Upload Permissions

Does Lawful Access to a Research Paper Mean You May Upload It to Any AI Service?

Lawful access to a research paper does not automatically authorize uploading it to any AI service. Reading, downloading, reproducing, and transmitting scholarly content may involve different rights and contractual conditions. Researchers should examine applicable licenses, copyright exceptions, institutional agreements, and AI service data practices before processing protected publications.

PART 0D · SUBPART 60 · GUIDE 829

Uploading Copyrighted Research Papers to AI

Can Researchers Upload Copyrighted Research Papers to an AI Tool?

Researchers may be permitted to upload copyrighted research papers to AI tools, but lawful access to an article does not automatically authorize every form of AI processing. Publisher licenses, contractual terms, copyright exceptions, institutional policies, and data handling arrangements may affect what is permitted. Researchers should verify these conditions before uploading protected documents.

PART 0D · SUBPART 60 · GUIDE 828

AI-Generated Images and Copyright Similarity

Can AI-Generated Images Resemble Copyrighted Images?

AI-generated images may resemble copyrighted photographs, illustrations, diagrams, and other visual works. However, similarity alone does not establish infringement. Researchers should distinguish unprotected ideas and artistic styles from protected expression, examine distinctive visual overlap, and verify relevant rights before using AI-generated graphics in scholarly publications.

PART 0D · SUBPART 60 · GUIDE 827

AI Reproducing Published Research Articles

Can Generative AI Reproduce Parts of Published Research Articles?

Generative AI can sometimes reproduce passages from published research articles, particularly when models memorize text or retrieve source documents. However, not every similarity indicates copying or copyright infringement. Researchers should verify distinctive passages, distinguish source-supported summaries from reproduction, and respect applicable attribution and copyright requirements.

PART 0D · SUBPART 60 · GUIDE 826

Publishing AI-Generated Text and Copyright

Can Researchers Publish AI-Generated Text Without Copyright Concerns?

Researchers may be permitted to publish AI-generated text, but publication is not automatically free from copyright concerns. Third-party protected expression, uncertain copyright ownership, service agreements, and publisher requirements may affect lawful and responsible use. Researchers should verify substantive content, examine relevant rights, and retain accountability for the final manuscript.

PART 0D · SUBPART 60 · GUIDE 825

Copyrighted Material in AI-Generated Output

Can AI-Generated Output Contain Copyrighted Material?

AI-generated output can contain material resembling or reproducing copyrighted works, including published text, code, images, and other protected expression. Researchers should distinguish ordinary similarities from potentially infringing reproduction, verify suspicious passages against original sources, and consider permissions, exceptions, and scholarly attribution requirements before publication.

PART 0D · SUBPART 60 · GUIDE 824

Copyright Protection for AI-Generated Research Text

Can AI-Generated Research Text Be Copyrighted?

AI-generated research text may not qualify for copyright protection in jurisdictions requiring human authorship, although original human-written, revised, or creatively arranged elements may be protected. Researchers should distinguish copyright eligibility from permission to use AI output, scholarly authorship, and compliance with publisher policies.

PART 0D · SUBPART 60 · GUIDE 823

Ownership of AI-Generated Research Text

Who Owns AI-Generated Research Text?

Ownership of AI-generated research text depends on applicable copyright law, the extent of human creative contribution, contractual terms, and institutional arrangements. Researchers should distinguish permission to use AI output from copyright protection and academic authorship, which involve different legal and scholarly responsibilities.

PART 0D · SUBPART 59 · GUIDE 822

AI Training-Data Inequality and Local Research Decisions

How Can Researchers Use AI Without Letting Global Training-Data Inequalities Dictate Local Research Decisions?

Researchers can use generative AI without allowing global training-data inequalities to determine local research decisions by treating AI outputs as provisional suggestions, examining their contextual assumptions, verifying recommendations against relevant evidence, and retaining responsibility for theoretical, methodological, and interpretive choices.

PART 0D · SUBPART 59 · GUIDE 821

AI and Equitable Access to Research Support

Could AI Reduce Some Inequalities in Access to Research Support?

Generative AI could reduce some inequalities in research support by making preliminary methodological explanations, programming guidance, writing assistance, and research planning resources more accessible. However, meaningful improvements depend on output reliability, researcher competence, connectivity, language support, and access to independent verification.

PART 0D · SUBPART 59 · GUIDE 820

AI and Inequality Between Research Institutions

Could Reliance on AI Create New Inequalities Between Researchers or Research Institutions?

Reliance on generative AI may create new inequalities between researchers and institutions when access to advanced models, computational infrastructure, specialist expertise, and reliable research support is uneven. Although AI can reduce some barriers, its benefits depend on institutional capacity, research governance, and whether technological advantages translate into stronger scientific evidence.

PART 0D · SUBPART 59 · GUIDE 819

AI Methodological Support for Researchers

Can Generative AI Help Researchers With Limited Access to Methodological Support?

Generative AI can help researchers with limited access to methodological support by explaining research concepts, comparing designs, identifying analytical assumptions, and preparing questions for specialist consultation. Its usefulness depends on independent verification, contextual suitability, and the researcher's ability to distinguish educational assistance from validated methodological expertise.

PART 0D · SUBPART 59 · GUIDE 818

Western Research Assumptions in Generative AI

Can Generative AI Impose Western Research Assumptions on Other Contexts?

Generative AI may reproduce Western-centric research assumptions when recommending theories, methods, measurements, or interpretations for other cultural and institutional settings. Researchers should critically evaluate whether AI-generated advice reflects the research context, respects locally relevant knowledge, and preserves scientific rigor without treating dominant academic conventions as universal.

PART 0D · SUBPART 59 · GUIDE 817

AI Research Methods in Resource-Limited Settings

Can Generative AI Recommend Research Methods That Are Unrealistic in Resource-Limited Settings?

Generative AI may recommend research designs that require unavailable datasets, expensive software, extensive funding, or institutional infrastructure. Researchers should evaluate methodological appropriateness and feasibility separately, adapting research questions and procedures without making unsupported compromises in scientific rigor.

PART 0D · SUBPART 59 · GUIDE 816

AI Misunderstanding Local Research Systems

Can Generative AI Misunderstand Local Research Systems, Institutions, or Regulations?

Generative AI may misunderstand local research systems by applying unfamiliar institutional procedures, outdated regulations, or assumptions drawn from other countries. Researchers should verify consequential advice against authoritative local sources and distinguish general methodological guidance from jurisdiction-specific requirements.

PART 0D · SUBPART 59 · GUIDE 815

AI Bias Toward Research From Highly Visible Countries

Can Generative AI Overrecommend Research From Highly Visible Countries?

Generative AI may overrecommend research from highly visible countries because of uneven publication coverage, citation networks, language representation, and retrieval ranking. Researchers should assess geographic patterns against the relevance and availability of the literature rather than assuming that frequently recommended countries produce the most appropriate evidence.

PART 0D · SUBPART 59 · GUIDE 814

AI Overlooking Local Research Literature

Can Generative AI Overlook Local Research Literature?

Generative AI may overlook local research literature because regional journals, institutional repositories, non-English publications, and locally relevant reports are unevenly represented in its accessible information. Researchers should verify AI-generated literature recommendations through independent searches and avoid interpreting missing AI results as evidence that relevant studies do not exist.

PART 0D · SUBPART 59 · GUIDE 813

AI Research Advice for Underrepresented Populations

Can AI Give Poorer Research Advice for Underrepresented Countries or Populations?

Generative AI may provide less appropriate research advice for underrepresented countries or populations when its recommendations rely on dominant scholarly contexts, incomplete knowledge of local institutions, or unrealistic assumptions about research resources. Researchers should evaluate advice against local requirements, methodological evidence, and the practical conditions of their study.

PART 0D · SUBPART 58 · GUIDE 812

Reporting AI Fairness Limitations

How Should AI Fairness Limitations Be Reported in Research?

Researchers should report AI fairness limitations by explaining which populations and fairness criteria were evaluated, what disparities were observed, and how those limitations affect the interpretation or use of research findings. Transparent reporting distinguishes demonstrated problems from untested risks and avoids presenting partial validation as proof that an AI system is universally fair.

PART 0D · SUBPART 58 · GUIDE 811

Choosing AI Fairness Definitions in Research

Who Should Decide Which Definition of AI Fairness Is Appropriate for a Research Study?

Selecting an AI fairness definition is not exclusively a technical decision. Researchers should justify fairness criteria through scientific objectives, affected stakeholder perspectives, ethical obligations, and the consequences of errors. Transparent governance is essential when competing definitions lead to different conclusions about whether an AI system is fair.

PART 0D · SUBPART 58 · GUIDE 810

Conflicting AI Fairness Metrics

Can Different AI Fairness Metrics Conflict With One Another?

AI fairness metrics can conflict because they measure different properties of predictions and decisions. Under certain conditions, a model cannot simultaneously satisfy calibration, equalized odds, and other fairness requirements. Researchers should understand these mathematical trade-offs, justify the criteria appropriate to their study, and report unresolved disparities transparently.

PART 0D · SUBPART 58 · GUIDE 809

Using AI With Known Population Performance Limitations

Should Researchers Use an AI Tool Known to Perform Worse for the Population Being Studied?

Using an AI tool known to perform worse for the study population may undermine research validity or create unequal consequences. Researchers should evaluate whether the tool meets task-specific requirements, whether limitations can be mitigated, and whether a more suitable alternative is available before relying on its outputs.

PART 0D · SUBPART 58 · GUIDE 808

Evaluating Acceptable AI Performance Differences

How Do Researchers Decide Whether AI Performance Differences Between Groups Are Scientifically Acceptable?

Researchers cannot determine whether AI performance differences between groups are acceptable from statistical significance or accuracy alone. A defensible assessment considers measurement validity, uncertainty, error consequences, predefined performance requirements, and the ethical implications of unequal outcomes.

PART 0D · SUBPART 58 · GUIDE 807

Biased AI Tools and Research Conclusions

Can a Biased AI Tool Produce Biased Research Conclusions?

Biased AI tools can influence research conclusions when systematic distortions affect evidence selection, measurement, classification, analysis, or interpretation. Researchers should trace AI-generated decisions through the workflow, evaluate whether errors materially change findings, and qualify conclusions that exceed the available evidence.

PART 0D · SUBPART 58 · GUIDE 806

Evaluating AI Performance Across Population Subgroups

Should Researchers Check AI Performance Across Relevant Population Subgroups?

Researchers should evaluate AI performance across relevant population subgroups when differences could affect scientific conclusions or consequential decisions. Subgroup analysis can reveal errors concealed by aggregate metrics, but its design must account for construct validity, sample size, uncertainty, and the consequences of observed disparities.

PART 0D · SUBPART 58 · GUIDE 805

AI Accuracy and Minority Subgroup Performance

Can an AI Tool Be Accurate Overall but Perform Poorly for a Minority Subgroup?

An AI tool may achieve excellent overall accuracy while performing poorly for a minority subgroup because aggregate metrics are dominated by larger populations. Researchers should examine subgroup-specific errors, sample sizes, calibration, and the consequences of unequal performance before relying on overall accuracy as evidence of reliability.

PART 0D · SUBPART 58 · GUIDE 804

Equal AI Performance and Fairness

Is Equal AI Performance Across Groups Enough to Establish Fairness?

Equal AI performance across groups may indicate that a model meets a particular evaluation criterion, but it does not establish fairness in every relevant sense. Researchers must also examine measurement validity, error consequences, access to benefits, and the social and institutional contexts in which AI outputs are used.

PART 0D · SUBPART 58 · GUIDE 803

AI Fairness in Scientific Research

What Does Fairness Mean When AI Is Used in Research?

AI fairness in research concerns whether AI-assisted methods, predictions, classifications, and decisions treat relevant populations appropriately and avoid unjustified disparities. Fairness cannot be established through accuracy alone and requires explicit consideration of scientific objectives, error consequences, population differences, and competing ethical criteria.

PART 0D · SUBPART 57 · GUIDE 802

Human Review and AI Bias in Research

Can Human Review Completely Eliminate AI Bias From Research?

Human review can identify and reduce some forms of AI bias, but it cannot guarantee that research outputs are unbiased. Reviewers may share the same assumptions as AI systems, overlook systematic errors, or become influenced by plausible recommendations. Effective oversight requires independent evidence, appropriate evaluation criteria, and meaningful authority to challenge AI-generated decisions.

PART 0D · SUBPART 57 · GUIDE 801

Testing AI Research Workflows for Bias

How Do Researchers Test an AI-Assisted Research Workflow for Bias?

Testing an AI-assisted research workflow for bias requires identifying where AI influences research decisions, defining relevant risks, and evaluating outputs against suitable evidence. Researchers can use representative samples, matched prompts, subgroup analyses, and sensitivity testing to determine whether systematic distortions threaten research validity.

PART 0D · SUBPART 57 · GUIDE 800

AI Bias Against Marginalized and Local Knowledge

Can AI Make Marginalized or Locally Specific Knowledge Less Visible in Research?

Generative AI may make marginalized and locally specific knowledge less visible when its outputs favor dominant languages, highly indexed publications, and familiar theoretical frameworks. Researchers should evaluate whose knowledge is represented, identify relevant omissions, and preserve context-specific interpretations without compromising scholarly rigor.

PART 0D · SUBPART 57 · GUIDE 799

AI Bias Toward Dominant Scientific Theories

Can AI Favor Dominant Theories or Mainstream Scientific Perspectives?

Generative AI may favor dominant scientific theories because prominent frameworks are extensively represented in scholarly material and AI retrieval systems. Researchers should evaluate theoretical recommendations against the research question, competing explanations, and relevant local knowledge rather than treating popularity as evidence of suitability.

PART 0D · SUBPART 57 · GUIDE 798

AI Bias in Scientific and Academic Writing

Can AI Bias Affect Scientific and Academic Writing?

Generative AI can introduce bias into scientific and academic writing by favoring familiar theories, altering uncertainty, standardizing culturally specific expressions, or selectively emphasizing evidence. Researchers should verify AI-assisted revisions against their original arguments, sources, and findings.

PART 0D · SUBPART 57 · GUIDE 797

AI Bias in Research Prediction Models

Can AI Bias Affect Prediction Models Used in Research?

AI bias can affect prediction models by producing systematic errors, unequal performance across populations, or misleading estimates. Researchers should examine data quality, model calibration, subgroup performance, and deployment conditions before relying on predictions in scientific analysis.