For Which Sciences Is AIIM Most Important?

Getting your Trinity Audio player ready...

AI Internet-Meritocracy (AIIM) is potentially most important for sciences in which valuable work is inexpensive to perform but difficult for conventional institutions to recognize, classify, or fund. These include fundamental mathematics, theoretical science, interdisciplinary and independent research, scientific software and data infrastructure, replication, negative results, and research addressing populations with little purchasing or political power.

AIIM may benefit every scientific discipline, but its comparative advantage is not equally large everywhere. A conventional grant can already work reasonably well for a recognized laboratory conducting a clearly defined, equipment-intensive project. AIIM becomes more distinctive when scientific value is distributed across many people and outputs, appears gradually, crosses institutional boundaries, or becomes visible only after the work has been published.

A Practical Ranking

Priority	Scientific area	Why AIIM is especially relevant
1	Fundamental mathematics and theoretical science	Important work may require mainly researcher time, while its value can be difficult to predict or evaluate through short proposals
2	Independent, unconventional, and interdisciplinary research	Researchers can fall outside established departments, grant categories, and professional networks
3	Research software, datasets, and scientific infrastructure	These outputs support many later discoveries but are often treated as auxiliary work
4	Replication, verification, and negative results	They create public value but provide weak conventional career and publication incentives
5	Neglected-disease and low-commercial-value research	Social importance can be high even when neither markets nor wealthy institutions provide adequate incentives
6	Research in poorly funded countries	Scientific talent and locally relevant problems are distributed more widely than research funding
7	Politically sensitive or institutionally unfashionable research	Distributed evaluation may reduce dependence on a single political, corporate, or academic authority

This is not a ranking of which sciences are most valuable. It is a ranking of where the difference between scientific value and conventional fundability may be greatest.

Fundamental Mathematics and Theoretical Science

Fundamental mathematics is probably the clearest case for AIIM.

A mathematician may need little more than time, access to literature, and computing resources. Nevertheless, evaluating an unfamiliar mathematical framework can require substantial specialist attention. The work may be too broad for one paper, too unconventional for an established grant category, or too early to have accumulated citations.

Conventional project funding asks researchers to describe:

• what they will discover;
• how they will discover it;
• how long discovery will take;
• what measurable impact it will produce.

These questions are often poorly matched to fundamental research. A mathematician cannot reliably schedule a theorem, and a genuinely new definition may initially appear less impressive than a large project built from familiar concepts.

AIIM reverses part of this process. Instead of paying mainly for a persuasive prediction, it can evaluate attributable public output that already exists: definitions, proofs, formalizations, monographs, software, explanatory work, and the dependencies that later research has on them.

This is why AIIM is closely related to the problem discussed in Why Fundamental Mathematics Cannot Be Funded Like a Startup and Why Long Mathematical Monographs Struggle Under Modern Evaluation Systems.

AIIM is not needed because mathematics receives no institutional support. The US National Science Foundation, for example, continues to finance major mathematical-sciences institutes. The problem is narrower: institutional programmes cannot reliably inspect every independent, highly specialized, or structurally unfamiliar contribution.

Independent and Interdisciplinary Research

AIIM may be even more important for a valuable researcher than for a particular discipline.

Conventional science funding is normally routed through universities, research institutes, national agencies, and predefined programmes. This creates several filters before the scientific content is evaluated:

1. The researcher must belong to an eligible organization.
2. The project must fit an administrative category.
3. The application must reach suitable reviewers.
4. The proposed work must appear credible before its results exist.
5. The institution must be willing to administer the award.

An independent researcher can fail at the first step even when the published work is substantial. Interdisciplinary research can fail because no individual committee considers it central to its remit.

AIIM can evaluate researchers and outputs across institutional and disciplinary boundaries. In principle, it can consider a mathematical theory, its implementation in software, its applications in another field, and the explanatory work needed for adoption as connected contributions rather than unrelated résumé entries.

This makes AIIM especially relevant to orphan science: important work without a clear institutional sponsor. See The Orphan Science Problem and Research Before Consensus.

Research Software, Data, and Scientific Infrastructure

Modern science depends on code, databases, libraries, benchmarks, documentation, standards, and maintenance. Yet these outputs often remain less visible than papers that use them.

A software maintainer may:

• fix a defect affecting hundreds of studies;
• preserve compatibility with new hardware;
• document an undocumented method;
• curate a dataset;
• review contributions;
• answer technical questions;
• keep an essential service operational.

None of these actions necessarily produces a conventional research paper. Nevertheless, removing the software or dataset could damage an entire research community.

AIIM is well suited to such work because dependency relationships are partly observable. Publications, repositories, packages, datasets, citations, imports, reuse, and documented applications can provide evidence that later work depends on an earlier contribution.

The European Open Science Cloud similarly treats reusable data, tools, and services as research infrastructure intended to increase productivity, reproducibility, and trust. UNESCO’s Recommendation on Open Science also calls for investment in platforms and infrastructure, including support for scientists in low- and middle-income countries.

AIIM could go further by rewarding the people who create and maintain infrastructure continuously rather than waiting for them to package maintenance work as a new project proposal.

Related articles include Measuring the Impact of Research Software and Mathematical Libraries and How Dependency Graphs Can Reveal Hidden Scientific Contributors.

Replication, Verification, and Negative Results

Replication is essential to reliable science, but the person who confirms or refutes an earlier result often receives less recognition than the person who made the original claim.

The same asymmetry affects:

• reproducing computational results;
• checking proofs and datasets;
• testing robustness under alternative assumptions;
• publishing failed experiments;
• documenting methods that do not work;
• correcting errors without producing a dramatic new conclusion.

The US National Academies distinguishes reproducibility, obtaining consistent computational results from the same inputs and procedures, from replicability, obtaining consistent results in a new study addressing the same question. Its recommendations explicitly involve not only researchers and journals but also institutions and funders.

AIIM can reward verification as an independent scientific output. A careful failed replication could receive value because it prevents other researchers from relying on an unreliable claim. A negative result could receive value because it prevents many laboratories from repeating an unproductive approach.

This is a major structural advantage: AIIM does not need to pretend that every valuable contribution is a novel positive discovery.

See Why Replication Should Be a Paid Scientific Profession and Can Failed Experiments Be Valuable Public Goods?.

Neglected Diseases and Other Low-Market-Value Research

Some research is neglected not because it lacks scientific or humanitarian importance, but because the people who need its results cannot create a sufficiently profitable market.

This is particularly visible in neglected tropical diseases. The World Health Organization reported that low-income countries received only 0.2% of grant funding from major international health-research funders in the dataset it analyzed, while neglected tropical diseases received approximately 0.6% of the examined funding. WHO has also noted that limited commercial incentives contribute to gaps in the development of treatments for neglected diseases.

The WHO’s 2025 global report further said that official development assistance for neglected tropical-disease programmes decreased by 41% between 2018 and 2023.

AIIM cannot manufacture unlimited money, but it can improve allocation within a donated global fund. Instead of asking whether research serves a profitable market or a politically powerful constituency, the system can ask what portion of available resources the work merits based on its expected and demonstrated scientific contribution.

This principle also applies to:

• rare diseases;
• diseases concentrated in poorer countries;
• low-cost public-health interventions;
• local environmental hazards;
• agricultural problems affecting small populations;
• language technologies for less commercially valuable languages.

Research in Poorly Funded Countries

Scientific ability is global, but research funding is geographically concentrated.

UNESCO reported that G20 countries accounted for roughly nine-tenths of global research expenditure, researchers, publications, and patents in the period covered by its 2021 Science Report. It also reported that 80% of countries invested less than 1% of GDP in research and development.

These disparities can become self-reinforcing:

• countries with less funding produce fewer publications;
• lower publication volume reduces international visibility;
• low visibility makes future grants harder to obtain;
• researchers migrate or leave science;
• locally important questions remain understudied.

AIIM can separate a scientist’s evaluation from the wealth of the scientist’s institution or country. A global AIIM fund could evaluate public work using common procedures and transfer support directly to researchers, subject to legal, identity, and anti-fraud safeguards.

This does not mean that an algorithm automatically eliminates geographic bias. Training data, publication access, language coverage, and digital visibility can reproduce existing inequalities. AIIM therefore needs multilingual evidence collection, transparent explanations, appeals, human voting, and adversarial testing.

Social Sciences and Politically Sensitive Research

Social sciences are important candidates for AIIM, although they present greater evaluation difficulties than formal mathematics or openly verifiable software.

Research about government, inequality, religion, institutional failure, conflict, or public policy may encounter pressure from political authorities, donors, universities, or ideological groups. A decentralized fund can reduce dependence on any one gatekeeper.

However, AI evaluation in these fields must distinguish among:

• empirical evidence;
• interpretation;
• normative judgment;
• political advocacy;
• methodological quality;
• popularity.

An AI system could otherwise mistake ideological agreement, media attention, or textual confidence for scientific merit. Human oversight and plural evaluation are therefore especially important.

AIIM should not be presented as an infallible political referee. Its more defensible role is to make funding decisions more distributed, auditable, contestable, and evidence-based than decisions controlled by a single authority.

Where AIIM Has a Smaller Initial Advantage

AIIM may initially be less transformative for research requiring extremely expensive physical infrastructure, such as:

• particle accelerators;
• space telescopes;
• national biobanks;
• large clinical trials;
• research reactors;
• specialized fabrication facilities.

Such projects require procurement, safety regulation, long-term institutional responsibility, coordinated staffing, and legal accountability. Evaluating individual merit does not by itself build a telescope or operate a clinical trial.

Even here, AIIM could reward researchers, software developers, data curators, reviewers, and infrastructure maintainers. But it should complement rather than immediately replace capable institutions.

The distinction is important:

AIIM is strongest at allocating recognition and researcher-level support. It does not automatically replace organizations needed to own equipment, employ personnel, protect participants, or manage physical hazards.

As AIIM develops, it could also allocate funding to teams and infrastructures. That extension would require additional governance mechanisms beyond individual salary recommendations.

The General Rule

AIIM is most important when several conditions coincide:

• valuable work can be publicly demonstrated;
• conventional funding depends heavily on status or affiliation;
• the result is difficult to predict in advance;
• contributions are distributed among many people;
• citations capture only part of the value;
• commercial incentives are weak;
• the research crosses disciplinary or national boundaries;
• the cost of evaluating a contribution is large relative to the cost of producing it.

By this test, fundamental mathematics, theoretical research, independent scholarship, open-source scientific software, research infrastructure, replication, and neglected global problems should be among AIIM’s earliest priorities.

AIIM Should Not Permanently Choose One “Best” Science

AIIM should not hard-code a permanent disciplinary hierarchy. That would reproduce one of the defects it is intended to reduce.

The appropriate allocation can change as:

• new evidence appears;
• research dependencies become visible;
• urgent global problems emerge;
• fields become overfunded or underfunded;
• researchers produce new work;
• contributors review and challenge evaluations.

The purpose of AI Internet-Meritocracy is not to declare mathematics, medicine, physics, or any other discipline universally superior. Its purpose is to distribute available funding according to the merit of actual contributors while making the decision open to inspection, challenge, and improvement.

Conclusion

AIIM is most important not necessarily for the largest sciences, but for the least accurately rewarded scientific work.

Its greatest potential lies in fields where a researcher can create enormous public value without commanding a large laboratory, belonging to a prestigious institution, promising an immediately marketable product, or fitting an established funding programme.

That makes fundamental and independent research the clearest starting point. But the same architecture could eventually support a much broader scientific ecosystem—one that rewards proofs, experiments, replications, software, data, maintenance, criticism, and communication as distinct but interconnected contributions.