AI Exposure and AI Resilience: A Two-Dimensional Assessment Framework for Software and Software-Based Business Model
Paul Darius Mandl, Peter Mandl, Martin Häusl
cs.AI
Quick verdict
A fully specified but unvalidated two-dimensional AI exposure/resilience scoring framework for software businesses; formal logic sound, evidence absent.
Summary
The paper introduces AI-ER, a two-dimensional framework for assessing how artificial intelligence affects software and software-based business models. AI exposure captures the pressure AI places on a company's value proposition, margins, and customer access; AI resilience captures the capacity to absorb that pressure, adapt, and use AI economically. From four recurring impact mechanisms (substitution, compression, bundling, re-intermediation) the authors derive four numerical exposure metrics plus a categorical business-model modulator, and five resilience metrics, scored on five-point scales. A rule-based, non-compensatory score logic combines them: exposure uses the maximum of substitutability and replicability plus a threshold indicator and the modulator; resilience uses the minimum of protective positions and adaptability plus an indicator and an economic-viability penalty. A confidence model combines five evidence-quality properties with inter-run rating agreement, taking the weaker of the two. An outside-in assessment uses three independent runs; a single inside-in run refines it with internal evidence. An illustrative fictitious company yields E=5, R=4 ('AI-Ready Compounder'). The framework is internally coherent and fully formalized, but entirely unvalidated: no real case, no reliability data, no longitudinal or sensitivity results, and the promised metric scale anchors are not printed; no code or data is available.
Review complete
Assessment at a glance
5
risk flags identified
● high 1
● medium 2
● low 2
5 strengths · 5 weaknesses · 5 author questions
AI text signal = Binoculars zero-shot detector score over 20 sampled passages (Fast-DetectGPT co-reported) — absence does not prove human authorship, and a flag does not imply misconduct.
Levels reflect verification complexity — not a quality score. Evolve never rates your paper.
Key claims under review
- AI exposure (external pressure for change) and AI resilience (capacity to absorb and adapt) are distinct dimensions that must be assessed separately, since high exposure can coexist with high resilience.
- Nine metrics plus a categorical business-model modulator, derived from four impact mechanisms (substitution, compression, bundling, re-intermediation), suffice to operationalize both dimensions for software-based businesses.
- A non-compensatory score logic — maximum for exposure core drivers, minimum for resilience core drivers, threshold indicators, and bounded modulators — keeps critical attack paths and weaknesses visible where averages would hide them.
- Explicit modeling of evidence quality and inter-run agreement separates the assessment result from its reliability, so weakly supported ratings remain visible instead of being concealed.
- The framework is explicitly a proposal: empirical validation via case studies, inter-rater reliability, longitudinal tests, and sensitivity analysis is required and has not yet been performed.
Strengths
Clean separation of two dimensions with full formalization
Secs. I and V-VI: exposure (external pressure) and resilience (response capacity) are kept separate and never collapsed into one grade; Eqs. (1)-(15), the symbol table (Table V) and the parameter table (Table VI) make the entire score logic deterministic and auditable, and the worked example's numbers (E=5, R=4, Q=0.8, A≈0.83) all verify by hand.
Non-compensatory design applied consistently
Max over exposure core drivers (Eq. 5) preserves the stronger attack path; min over resilience core drivers (Eq. 8) keeps the weakest condition visible; confidence takes min{Qij, Aij} (Eq. 13) so strong evidence cannot offset poor inter-run agreement — a coherent MCDA stance explicitly grounded in Roy [23] and Munda [24].
Honesty infrastructure for weak evidence
Configuration transparency including provenance
Pre-registered validation program
Weaknesses
Zero empirical content
Promised scale anchors are absent
Uncalibrated structural choices with saturation
Circularity risk in the validation design
Gaming and adversarial robustness unaddressed
Full Review Report
$1.99 · one-timeUnlock the complete preliminary analysis Evolve produced for this paper.
Report preview
Review: AI Exposure and AI Resilience (AI-ER)
The remaining sections cover Methodology Assessment, Statistical Evidence, Results & Reproducibility, Limitations & Risks, and Verdict & Recommendations.
- All 5 strengths & 5 weaknesses, analyzed in detail
- 5 risk flags with severity, section & reviewer questions
- Methods & statistical evidence assessment
- 5 questions a peer reviewer would ask
- Full six-dimension evaluation report & recommendations
One-time purchase · this paper only
One-time purchase, this paper only. AI-generated preliminary analysis — not peer review.
Frequently asked questions
What is the AI-ER framework?
AI-ER (AI Exposure and AI Resilience) is a proposed two-dimensional assessment method for software and software-based businesses. AI exposure is scored 1-5 for how much change pressure AI creates for a company's value proposition, margins, and customer access; AI resilience is scored 1-5 for the company's capacity to absorb that pressure and adapt. The two scores are shown together as quadrants — Defended Niche, AI-Ready Compounder, Rebuilding Required, Acute Threat — rather than merged into a single grade.
Has the AI-ER framework been validated on real companies?
No. The paper is a conceptual proposal: it formalizes metrics, score rules, and a confidence model, and sketches a validation plan (comparative case studies, inter-rater reliability via Krippendorff's alpha or Cohen's kappa, longitudinal tracking, sensitivity analysis), but reports no real-company assessment, no reliability data, and no predictive results.
How are the AI-ER scores computed?
Each of nine metrics is rated 1-5 from documented evidence. Exposure = max(substitutability, replicability) + 1 if competitive dynamics or customer access is at least 4, plus or minus 1 from a business-model modulator. Resilience = min(protective positions, adaptability) + 1 if technical AI maturity and implementation capability are both at least 4, minus 1 if economic viability is 2 or below. Results are clamped to 1-5. Confidence blends evidence quality and agreement across independent assessment runs, keeping whichever is weaker.
Is code or data available for AI-ER?
No. The paper specifies the rules in enough detail to implement them — equations (1)-(15), a symbol table, and a parameter table with default values — but no software implementation, dataset, or real case evidence is released.
About this review
This is an AI-generated preliminary review by Evolve.Pub, produced from the openly available text of the paper. It is not peer review and does not represent the authors' views. The canonical record of the paper is at the publisher link above.
- PDF extraction flattens mathematical notation (e.g., Eq. (5) renders as 'maxxsub i , xrep i' with lost parentheses); all equations were reconstructable from the Sec. VI text and the Table V notation, so no equation-level judgment is at risk.
- Figures 1-3 appear as flattened inline label runs rather than structured captions; their content appears complete and no figure-only information was needed for this review.
Go deeper on this paper
Wrote a paper? Stress-test it before reviewers do.
Upload your manuscript for the same structured scrutiny — reasoning, statistics, evidence, figures, and reproducibility.