How We Evaluate AI Innovation
The International AI Awards evaluates AI projects on five principles: innovation and originality, technical execution, practical value and impact, reliability and responsible AI, and evidence and credibility. Reviewers consider the project’s category, maturity and available evidence.
A Rigorous, Evidence-Informed Approach to AI Excellence
At the International AI Awards (INAI), we believe meaningful AI innovation deserves more than an impressive presentation.
Our evaluation approach considers what a project claims to achieve, how it works, the problems it addresses, the evidence supporting its performance, and its potential to deliver meaningful value.
We recognise that AI innovation takes many forms, from independently developed machine learning models and early-stage prototypes to enterprise systems, creative applications, and technologies already in operational use.
Each submission is evaluated in the context of its category, purpose, maturity, and available evidence.
Our Five Evaluation Principles
1. Innovation & Originality
We examine the originality of the project’s approach, its contribution to solving a meaningful problem, and the extent to which it advances or meaningfully applies existing AI capabilities.
Thoughtful engineering, distinctive applications, and effective integration can represent meaningful innovation, including projects that build on existing foundation models.
2. Technical Execution
We consider the quality and completeness of the technical implementation, including system architecture, model design, functionality, and suitability for its intended purpose.
Where appropriate and accessible, reviewers may examine working demonstrations, technical documentation, software implementations, or source code.
3. Practical Value & Impact
We assess whether a project addresses a genuine need and demonstrates meaningful potential or measurable outcomes. Relevant evidence may include operational improvements, user adoption, productivity gains, cost reductions, research findings, educational benefits, or broader social impact.
Early-stage projects are considered in relation to their development stage.
4. Reliability & Responsible AI
We consider the reliability of AI-generated outputs and the project’s approach to uncertainty, errors, safety, privacy, and appropriate human oversight.
Reviewers may examine whether systems distinguish verified information from assumptions, acknowledge limitations, and implement safeguards appropriate to their intended applications.
5. Evidence & Credibility
Claims are assessed in relation to the supporting evidence available. Evidence may include independent references, published research, working prototypes, technical benchmarks, customer feedback, operational records, and documented results.
We distinguish independently verifiable findings from applicant-reported achievements. We value clear communication of limitations alongside demonstrated performance.
How We Assess Human Contribution and Future Potential
Reviewers look at the decisions people made: how they defined the problem, designed or adapted the system, directed the creative work, tested outputs, and responded to limitations. Those contributions can be meaningful at any team size.
We weigh future potential in context. A prototype can make a credible case through a working demonstration, early evidence, and a clear path to practical use. An established system can demonstrate its value through deployment, reliability, and measured outcomes.
For AI-generated or AI-assisted works, including visual art, music, video, and digital media, we evaluate both the artistic quality of the final creation and the depth of the creative process behind it.
Beyond originality, aesthetics, emotional expression, and technical execution, we consider the creator's artistic vision, thoughtful use of AI tools, prompt development, experimentation, iterative refinement, and critical decision-making throughout the creative journey.
We value works that demonstrate intentional artistic direction, meaningful human contribution, and a clear evolution from initial concept to final outcome. The sophistication of AI tools alone does not determine artistic merit; what matters is how creatively and purposefully they are used to achieve a compelling result.
Beyond the Application Form
Depending on the nature of the project and the materials available, our evaluation process may include the following methods.
-
Background and credibility review — Examining relevant company information, professional backgrounds, published materials, and independent references.
-
Product and demonstration testing — Interacting with accessible software or demonstrations to examine functionality and the consistency of AI behaviour.
-
Technical and model review — Reviewing available code, technical reports, evaluation methods, or performance metrics, and reproducing selected results where feasible.
-
Independent cross-checking — Comparing relevant project claims or outputs against authoritative datasets, published sources, or independent evidence.
-
Impact and maturity assessment — Considering the strength of deployment claims, user adoption, measurable outcomes, and project readiness.
-
Risk and limitations review — Examining potential weaknesses, unsupported assumptions, reliability concerns, and areas requiring further validation.
The methods applied depend on the project’s category, technical characteristics, accessibility, and submitted supporting materials.
Evidence and Transparency
We assess projects on demonstrated achievement. A promising prototype may be recognised for its originality and technical merit, while an established enterprise solution may be assessed on the strength of its implementation, reliability, and measurable impact.
For projects that use existing AI models or third-party technologies, we examine the contribution made through their development, application, integration, or improvement.
Limited supporting evidence may constrain what reviewers can independently verify. Early-stage and confidential projects remain eligible for consideration.
Human Judgment Remains Central
AI tools may support technical examination, information gathering, consistency checks, and analysis of submitted evidence. Reviewers consider AI-generated assessments alongside source materials, technical context, and identified limitations.
Appointed reviewers and competition organisers remain responsible for assessment and final award decisions.
Recognising Achievement at Every Stage
The International AI Awards seeks to recognise work that demonstrates genuine thought, technical effort, innovation, and meaningful potential. Our goal is to recognise contributions that advance the responsible development and practical application of artificial intelligence.
We value substance, evidence, and meaningful progress.
.png)