Civil Law and UAE: Shift from Argument to Computation in Legal Reasoning

1. Introduction

The phrase “shift from argument to computation in legal reasoning” describes a developing transformation in legal practice in which some parts of legal analysis move from purely narrative and argumentative methods toward data processing, structured rules, algorithms, probability, automated evidence analysis, decision trees, quantitative models and artificial intelligence.

In the UAE, this shift is visible particularly in:

electronic evidence;

digitally generated records;

automated court systems;

AI-assisted legal research;

smart forms;

Digital Economy Court proceedings;

algorithmic analysis of financial transactions;

electronic discovery;

automated document review; and

AI-assisted drafting.

However, it is important to state the legal position precisely:

UAE law has not replaced judicial argument with computerised decision-making.

Rather, computation is increasingly becoming a tool for organising, testing and evaluating legal material, while the legal decision remains subject to applicable law, evidence, judicial reasoning and human responsibility.

The DIFC Courts' 2023 guidance on generative AI expressly states that AI should assist legal submissions rather than replace the human decision-making integral to court proceedings. (DIFC Courts)

2. Meaning of “Argument to Computation”

Traditional legal reasoning can be represented as:

Facts → Evidence → Legal rules → Interpretation → Argument → Judicial conclusion

Computational legal reasoning introduces another layer:

Data → Structured facts → Algorithmic processing → Pattern/probability/model → Legal rule → Human verification → Judicial conclusion

For example, instead of manually reviewing 500,000 financial transactions, a computer can identify:

unusual payments;

duplicate invoices;

related-party transactions;

transactions outside ordinary ranges;

suspicious timing;

matching counterparties; and

unexplained financial flows.

The computer does not automatically determine liability.

It produces information that can become part of the evidential and argumentative process.

3. Why the UAE Is Relevant to This Shift

The UAE has developed several digital judicial and legal initiatives.

The Federal Evidence Law No. 35 of 2022 expressly recognises forms of electronic evidence including:

electronic instruments;

electronic signatures;

electronic seals;

emails;

modern communication methods;

electronic media; and

other electronic evidence.

It also gives qualifying formal electronic evidence the same evidentiary value as formal instruments. (UAE Legislation)

The DIFC has gone further in certain procedural areas.

Its Digital Economy Court rules provide for proceedings to make extensive use of information technology and allow smart forms and AI-driven decision-tree software to obtain information necessary for conducting and disposing of certain claims. (DIFC Courts)

This is a significant institutional example of computation entering legal procedure.

4. Traditional Legal Argument

Traditional civil-law reasoning generally depends upon:

1. Identification of facts

What happened?

2. Proof

What evidence establishes it?

3. Legal classification

Is the conduct:

breach?

negligence?

fraud?

unjust enrichment?

valid performance?

force majeure?

4. Interpretation

What does the relevant legal provision mean?

5. Application

How does the rule apply to the proven facts?

6. Reasoned conclusion

What remedy follows?

The process is fundamentally interpretive and argumentative.

5. Computational Legal Reasoning

Computational reasoning can assist at almost every stage.

Fact extraction

AI can extract:

dates;

parties;

amounts;

contractual clauses;

transactions;

communications.

Classification

Systems can classify documents into:

contracts;

invoices;

correspondence;

evidence;

privileged documents;

potentially relevant records.

Rule matching

A legal database can connect facts to:

statutory provisions;

regulations;

precedents;

contractual clauses.

Quantification

Software can calculate:

damages;

interest;

contractual penalties;

loss periods;

financial exposure;

probabilities;

valuation ranges.

Pattern detection

Algorithms can identify patterns that may not be immediately apparent to humans.

6. The Important Difference: Computation Is Not Judgment

This is the central legal principle.

A computer may calculate:

AED 8.7 million of transactions fall outside the identified pattern.

But it cannot automatically establish:

“The defendant is legally liable for AED 8.7 million.”

The court must still determine:

whether the data is authentic;

whether the dataset is complete;

whether the algorithm is reliable;

whether the identified pattern is legally relevant;

whether causation is established;

whether the defendant has an explanation;

whether the relevant legal standard is satisfied; and

what remedy the law permits.

Therefore:

Computation assists fact-finding; it does not automatically create legal liability.

7. At Least 6 UAE/DIFC Case Laws

Case 1 — Graciela Limited v Giacobbe [2014] DIFC CFI 027

Principle

The DIFC Court explained the civil standard of proof as the balance of probabilities.

The court must assess competing probabilities and determine whether the relevant event is more likely than not to have occurred. (DIFC Courts)

Relevance to computation

This provides a conceptual bridge between traditional evidence and computational analysis.

An algorithm may calculate:

probability distributions;

frequency;

statistical anomalies;

comparative likelihood.

But the legal standard remains a legal standard of proof, not merely a mathematical probability score.

For example:

Algorithm: transaction anomaly probability = 92%.

This does not automatically mean:

Legal fact proven = 92%.

The court must evaluate the evidence as a whole.

Case 2 — VTJ Ltd v Mohammed Ammar Al Hassan [2018] DIFC CA 009

This line of DIFC authority concerns the evaluation of evidence and factual inference.

Importance

Civil courts frequently have to draw inferences from multiple pieces of evidence rather than relying on one direct statement.

That makes the case useful for understanding computational legal reasoning because algorithms similarly operate through:

correlations;

patterns;

probabilities;

cumulative indicators.

Limitation

A computational correlation is not automatically a legal inference.

The judge must determine whether the inference is justified by admissible evidence and the applicable legal test.

Thus:

statistical correlation ≠ legal causation.

Case 3 — SBM Bank (Mauritius) Ltd v Renish Petrochem FZE [2022] DIFC CA 011

This case concerned a major fraud claim involving financial transactions.

The DIFC Court of Appeal dealt with a substantial judgment arising from alleged fraudulent conduct connected with payments under a facility agreement. (DIFC Courts)

Relevance to computational reasoning

Financial fraud litigation is particularly suitable for computational tools because large transactional datasets can be examined for:

payment sequences;

unusual transfers;

transaction timing;

counterparties;

account movements;

inconsistencies.

But the court's task remains legal and evidential.

The algorithm may reveal a pattern.

The judge must decide whether the pattern, together with admissible evidence, establishes fraud.

Principle

Data pattern → evidential inference → legal finding

rather than:

Data pattern → automatic liability.

Case 4 — Techteryx Ltd v Aria Commodities DMCC & Others [2025] DIFC DEC 001

This is particularly important because it was decided by the DIFC Digital Economy Court.

The dispute involved approximately USD 456 million said to represent reserves backing the TrueUSD stablecoin. The proceedings involved digital assets, tracing, proprietary claims and worldwide/freezing relief. (DIFC Courts)

Computational significance

Digital-asset disputes can require analysis of:

blockchain transactions;

wallet addresses;

transaction hashes;

movement of digital assets;

tracing;

timestamps;

cryptographic records;

linked transactions.

This produces a form of evidence that is inherently computational.

A blockchain investigator can potentially reconstruct:

Wallet A → Wallet B → Wallet C → exchange → Wallet D.

The legal question is then:

What legal consequence follows from that transaction history?

Principle

Computational traceability can establish an evidential pathway, but legal ownership and liability still require legal analysis.

This case therefore represents the transition from conventional documentary evidence toward digitally structured evidence.

Case 5 — Krystal Financial Consultants LLC v Nextgen Robopark Investment LLC [2025] DIFC CA 007

The DIFC Court of Appeal's 2026 judgment concerned the standard governing appellate review of an evaluative first-instance decision. The Court rejected an overly mechanical approach to appellate intervention and explained that the position is more nuanced than simply asking whether the first-instance decision was “plainly wrong.” (DIFC Courts)

Relevance

This case demonstrates an important limitation on computational legal reasoning.

Legal reasoning often involves evaluative judgment.

A model may produce:

a probability;

a risk score;

a classification;

a ranking of evidence.

But a legal decision may require evaluation of:

credibility;

context;

competing explanations;

proportionality;

contractual purpose;

factual nuance.

These cannot necessarily be reduced to a single numerical output.

Principle

Legal evaluation is not always reducible to an algorithmic score.

Case 6 — Khaled Salem Musabeh Humad Al Mheiri v John Cameron [2025] DIFC CA 008

This recent DIFC Court of Appeal decision is particularly relevant to the relationship between evidence, reasoning and judicial explanation.

The Court emphasized the importance of adequate reasons, particularly where serious allegations such as fraud are determined. It explained that identifying the evidence, factual findings and reasoning process helps reduce the risk of error and allows the parties and appellate court to understand how the conclusion was reached. (DIFC Courts)

Relevance to computational reasoning

This is directly relevant to algorithmic decision-making.

Suppose an AI system produces:

“High probability of fraud.”

That output is insufficient by itself.

A legally accountable decision needs to explain:

which evidence was relied upon;

which facts were established;

which legal rule was applied;

how the evidence supported the factual finding; and

why the legal consequence followed.

Thus:

Explainability is a legal requirement of reasoning, not merely a technical feature of AI.

Case 7 — Fidel v Felecia & Faraz [2015] DIFC CA 002

This case concerned how the DIFC Court approaches questions of non-DIFC UAE law and evidence.

The Court held that it was not bound by a rigid English approach requiring every question of non-DIFC UAE law to be proved as foreign law through expert evidence. The DIFC Courts possess flexibility concerning evidentiary rules and may consider the circumstances and judicial expertise. (DIFC Courts)

Relevance

This demonstrates that legal reasoning is context-sensitive.

A computational system might operate through a fixed rule:

“UAE law = foreign law = expert evidence.”

But the actual judicial approach can be more nuanced.

Therefore, legal technology must encode:

jurisdiction;

applicable law;

judicial discretion;

procedural context;

evidential rules.

Case 8 — Al Buhaira National Insurance Company v Waleed Mohammad & Others [2026] DIFC CFI 010

This recent case illustrates another important limitation.

The court explained that contractual construction is not simply a literal exercise. The question involves what reasonable people with the relevant knowledge would have understood from the language used at the time of contracting. (DIFC Courts)

Computational relevance

A machine can identify:

words;

clauses;

definitions;

repeated expressions;

semantic similarities.

But contractual interpretation can require:

commercial context;

reasonable-person analysis;

contractual purpose;

surrounding circumstances.

Therefore:

semantic computation ≠ complete contractual interpretation.

9. Federal Evidence Law and Computational Evidence

Federal Decree-Law No. 35 of 2022 is extremely important.

The Evidence Law recognises electronic evidence and gives certain formal electronic evidence equivalent probative value to formal instruments where statutory conditions are met. (UAE Legislation)

This creates a legal environment where courts can increasingly work with:

system-generated records;

electronic transactions;

emails;

digital signatures;

electronic communications;

automated records;

databases.

This is a major foundation for computational legal reasoning.

10. From Documents to Data

Traditional evidence:

Contract.pdf

Modern evidence:

Contract + metadata + version history + email chain + access logs + digital signature + database record + transaction history.

The legal lawyer increasingly needs to understand data structures, not just documents.

For example:

Traditional question

“Did the defendant sign the contract?”

Computational evidence questions

Which device generated the signature?

What was the timestamp?

Was the document altered?

What IP address was used?

Was the signature certificate valid?

Was the document subsequently modified?

Does the system audit log confirm execution?

The legal argument becomes increasingly data-driven.

11. Smart Forms and Decision Trees

The DIFC Digital Economy Court Rules are particularly significant.

Rule 58.12 permits the Court to operate an electronic dynamic system using smart forms or AI-driven forms, including decision-tree software, to obtain information necessary for conducting and disposing of claims. (DIFC Courts)

This is a genuine example of computation entering procedural legal reasoning.

A decision tree may operate like:

Is there a written contract?

↓ Yes

Is there a jurisdiction clause?

↓ Yes

Does the clause cover the dispute?

↓ Yes

Is there an objection to jurisdiction?

↓ Yes

Proceed to jurisdictional analysis.

The system structures the issue.

But the ultimate legal determination may still require judicial assessment.

12. AI-Assisted Legal Research

AI can process enormous quantities of:

legislation;

case law;

contracts;

regulatory decisions;

evidence;

correspondence.

It can identify:

similar cases;

recurring legal concepts;

contradictory authorities;

relevant statutory provisions;

potential arguments.

This can change the lawyer's role from:

“Find and manually read everything.”

toward:

“Design the query, verify the results and evaluate legal significance.”

13. DIFC AI Guidance

The DIFC Courts issued Practical Guidance Note No. 2 of 2023 concerning large language models and generative AI in court proceedings.

The guidance specifically identifies risks including:

incorrect information;

misleading evidence;

confidentiality breaches;

intellectual-property issues;

data protection issues;

algorithmic bias.

It requires users to verify AI-generated material and stresses that AI should not replace human decision-making. (DIFC Courts)

This is highly relevant to the “argument → computation” thesis.

The UAE legal system is not saying:

“Let the machine decide.”

It is closer to:

“Use computation, but retain legal responsibility and human verification.”

14. Mathematical Damages

One of the clearest areas where computation is already legitimate is damages calculation.

Suppose:

Contract value = AED 10 million

Delay = 120 days

Daily contractual rate = AED 20,000

The computational calculation is:

120 × AED 20,000 = AED 2.4 million

But the court still has to determine:

whether delay occurred;

whether the defendant caused it;

whether the contractual clause applies;

whether the clause is enforceable;

whether the amount may be adjusted;

whether another statutory limitation applies.

Thus:

Arithmetic is computational; legal entitlement remains argumentative.

15. Probability and Civil Proof

Computational systems increasingly use probability.

For example:

70% likelihood of transaction being anomalous;

85% probability of document similarity;

95% probability two records came from the same source.

But civil law does not generally operate on the principle:

“AI probability above 51% = judgment for claimant.”

The legal standard must be translated into the applicable evidentiary framework.

Graciela v Giacobbe demonstrates the balance-of-probabilities standard, but the judicial determination remains a holistic assessment of evidence. (DIFC Courts)

16. Why Legal Reasoning Cannot Be Completely Computational

A. Legal language is ambiguous

Words such as:

reasonable;

fair;

material;

substantial;

excessive;

proportionate;

good faith;

require contextual interpretation.

B. Facts are incomplete

Legal disputes rarely provide a perfectly clean dataset.

Evidence may be:

missing;

contradictory;

manipulated;

incomplete;

unlawfully obtained.

C. Legal rules conflict

A case may involve:

contract law;

company law;

evidence;

public policy;

consumer protection;

data protection.

An algorithm must determine which rule has priority.

D. Values are involved

Courts sometimes balance:

contractual freedom;

fairness;

public policy;

proportionality;

commercial certainty.

These are not purely mathematical questions.

17. Computation and Judicial Discretion

A useful model is:

Level 1 — Automated

data collection;

document sorting;

arithmetic;

duplicate detection;

chronology creation.

Level 2 — Assisted

legal research;

case comparison;

anomaly detection;

probability analysis;

damages modelling.

Level 3 — Human legal judgment

interpretation;

credibility;

legal classification;

proportionality;

public policy;

final judgment.

The UAE's current direction is best understood as automation and assistance at Levels 1 and 2, rather than complete replacement of Level 3.

18. The “Computational Lawyer”

The emergence of computational legal reasoning may create a new professional role.

The lawyer increasingly needs to understand:

Legal skills

statutory interpretation;

precedent;

evidence;

procedure.

Data skills

databases;

data structures;

statistical reasoning;

data provenance.

AI skills

prompt design;

model limitations;

hallucination detection;

algorithmic bias;

explainability.

Verification skills

source checking;

authority validation;

version control;

jurisdiction checking.

The lawyer therefore becomes partly a legal-data interpreter.

19. Computational Reasoning and Evidence Reliability

A computational output should be evaluated through:

Input quality

Garbage in → garbage out.

Data completeness

Was the entire dataset considered?

Algorithm design

What assumptions were programmed?

Training data

What data was used to train the model?

Error rate

How often does the system produce false positives or false negatives?

Explainability

Can the result be understood?

Reproducibility

Can another expert reproduce the result?

Chain of custody

Can the origin and integrity of the digital evidence be established?

These questions are particularly important under the UAE's electronic-evidence framework.

20. Algorithmic Bias

Suppose an AI system predicts that a particular category of debtor is likely to default.

If the training dataset contains historical bias, the algorithm may reproduce it.

The legal danger is:

Historical pattern → algorithm → apparently objective score → discriminatory outcome.

The fact that the result is mathematical does not make it legally neutral.

The DIFC's AI guidance expressly warns about potential bias and inaccuracies and requires verification. (DIFC Courts)

21. Automation Bias

Another risk is automation bias.

This occurs when a human decision-maker gives excessive weight to a computer-generated conclusion simply because it appears objective.

Example:

AI risk score = 91%.

A judge or lawyer might subconsciously assume:

“The computer has already analysed the case.”

That would be problematic.

The correct approach is:

“Why did the system produce 91%, and does the underlying evidence legally establish the relevant fact?”

22. Explainability as a Legal Requirement

A computational legal system should ideally produce:

Input → Method → Processing → Output → Reasons

For example:

10,000 transactions
↓
identified 120 unusual transactions
↓
compared with historical transaction pattern
↓
17 transactions matched specified risk indicators
↓
human review
↓
legal conclusion

This is much safer than:

“AI detected fraud.”

The latter does not adequately explain the reasoning.

The Al Mheiri v Cameron decision's emphasis on identifying evidence, findings and reasoning is highly relevant to this principle. (DIFC Courts)

23. Computational Contract Interpretation

AI can analyse:

contractual definitions;

repeated terminology;

clause relationships;

cross-references;

drafting inconsistencies;

version changes.

But Al Buhaira National Insurance v Waleed Mohammad demonstrates that contractual construction involves understanding what reasonable parties would have understood in context, not merely literal textual processing. (DIFC Courts)

Therefore:

Natural-language processing is an aid to interpretation, not a substitute for legal construction.

24. Digital Assets and Computational Legal Reasoning

Techteryx v Aria Commodities provides an important illustration.

Digital assets create evidence that can be inherently computational.

A blockchain may provide:

immutable transaction records;

timestamps;

wallet addresses;

transaction hashes;

transfer history.

A legal dispute may therefore begin with computational reconstruction:

Asset → wallet → transfer → intermediary → exchange → destination.

The court then applies legal concepts such as:

ownership;

beneficial ownership;

tracing;

proprietary claims;

unjust enrichment;

fiduciary obligations;

injunctions.

Thus:

Blockchain computation → legal classification → judicial remedy.

25. Computation and Access to Justice

Computational tools can potentially reduce:

document-review costs;

research time;

repetitive procedural work;

administrative burdens.

The DIFC's digital procedure framework expressly seeks to use technology to improve efficiency and reduce the cost and environmental impact of proceedings. (DIFC Courts)

This can particularly benefit:

small claims;

repetitive commercial disputes;

document-heavy litigation;

digital asset disputes;

banking disputes.

But efficiency cannot override:

fairness;

due process;

confidentiality;

evidentiary reliability;

judicial independence.

26. From “Lawyer vs Lawyer” to “Model vs Model”

An emerging possibility is that litigation becomes partly computational.

For example:

Claimant's system

Predicts:

AED 15 million damages.

Defendant's system

Predicts:

AED 4 million damages.

Expert model

Produces:

AED 8–11 million range.

The court still has to decide:

which assumptions are correct;

which evidence is reliable;

what legal rule applies;

what losses are recoverable.

Therefore, computational litigation may create a new form of model-based advocacy.

27. Risks of the Shift

1. Black-box reasoning

Nobody understands why the model produced the result.

2. Hallucination

AI may invent:

cases;

statutory provisions;

quotations;

legal propositions.

3. Outdated law

The model may rely on repealed legislation.

This is particularly important in the UAE because the 2025 Civil Transactions Law replaced the 1985 Civil Transactions Law from 1 June 2026.

4. Jurisdictional confusion

A model may mix:

mainland UAE law;

DIFC law;

ADGM law;

English law.

5. Data bias

Historical datasets can reproduce past errors.

6. Privacy

Legal datasets may contain sensitive personal information.

7. Automation bias

Humans may over-trust computer outputs.

8. Loss of contextual judgment

A numerical model may overlook commercially or socially significant circumstances.

28. Safeguards for Computational Legal Reasoning

A UAE computational legal system should incorporate:

Human oversight

Final legal conclusions should remain subject to authorised human decision-making.

Source verification

Every legal proposition should be traceable to:

statute;

judgment;

regulation;

authoritative legal source.

Version control

The system must identify which law was applicable on the date of the transaction or event.

Jurisdiction control

The system must distinguish:

Federal UAE;

Emirate-level law;

DIFC;

ADGM;

other free zones.

Explainability

Every material computational conclusion should have an understandable reasoning trail.

Auditability

Inputs and outputs should be recorded.

Data security

Legal information must be protected.

Bias testing

Models should be periodically tested for systematic error.

29. Six Key Cases and Their Computational Significance

30. Mainland UAE vs DIFC

It is important not to treat the UAE as one uniform procedural system.

Mainland UAE

The Federal Evidence Law expressly recognises electronic evidence and provides rules concerning its evidentiary value. (UAE Legislation)

DIFC

The DIFC has gone further in procedural digitisation, including:

AI-driven smart forms;

Digital Economy Court procedures;

specific guidance concerning generative AI.

(DIFC Courts)

ADGM

ADGM has its own courts, evidence regulations and procedural framework. It should therefore be analysed separately from mainland UAE and DIFC. (ADGM)

31. Is UAE Civil Law Actually Moving From Argument to Computation?

The most accurate answer is:

Partially, but not completely.

The shift is occurring in the method of processing information, not in the fundamental source of legal authority.

The transformation can be expressed as:

Old model

Lawyer → documents → argument → judge

Emerging model

Data → algorithm → structured evidence → lawyer/judge → legal reasoning

Possible future model

Data → AI analysis → probability/model → human verification → legal argument → judicial decision

The final stage remains important.

32. Core Legal Principle

The central rule can be stated as:

Computation may increase the speed, scale and consistency of legal analysis, but numerical output does not itself constitute a legal conclusion.

For example:

Algorithm detects anomaly

≠

Fraud established

AI predicts breach

≠

Breach legally proven

Statistical correlation

≠

Causation established

Risk score

≠

Liability

Semantic similarity

≠

Correct statutory interpretation

33. Examination-Oriented Answer

Definition

The shift from argument to computation describes the increasing use of algorithms, data analytics, AI, automated evidence processing and quantitative models in legal reasoning.

UAE relevance

The shift is visible through:

electronic evidence under Federal Evidence Law No. 35 of 2022;

smart judicial systems;

Digital Economy Court;

AI-driven smart forms;

blockchain evidence;

AI-assisted legal research;

automated document analysis.

Key cases

Graciela Ltd v Giacobbe [2014] DIFC CFI 027 — balance of probabilities.

VTJ Ltd v Mohammed Ammar Al Hassan [2018] DIFC CA 009 — evidential inference.

SBM Bank v Renish [2022] DIFC CA 011 — financial fraud and evidence.

Techteryx v Aria Commodities [2025] DIFC DEC 001 — digital assets and computational tracing.

Krystal v Nextgen [2025] DIFC CA 007 — evaluative judicial decisions.

Al Mheiri v Cameron [2025] DIFC CA 008 — reasoned decision-making.

Fidel v Felecia & Faraz [2015] DIFC CA 002 — flexible evidentiary/legal methodology.

Al Buhaira National Insurance v Waleed Mohammad [2026] DIFC CFI 010 — contextual interpretation.

Formula

Data + Algorithm + Evidence + Human Verification + Legal Rule = Computational Legal Reasoning

34. Conclusion

The UAE is developing a legal environment in which computation increasingly assists legal reasoning, particularly in electronic evidence, financial analysis, digital assets, document review, smart judicial systems and AI-assisted legal work.

The development is especially visible in the DIFC. Its Digital Economy Court rules expressly contemplate AI-driven smart forms and decision-tree systems, while its 2023 AI guidance requires transparency, verification and caution against excessive reliance on generative AI. (DIFC Courts)

At the federal level, the Evidence Law's recognition of electronic evidence provides the statutory foundation for courts to deal with increasingly data-based disputes. (UAE Legislation)

The case law nevertheless demonstrates an important boundary. Graciela shows that probability operates within a legal standard of proof; Techteryx shows how digital-asset disputes can depend upon computational tracing; Al Mheiri emphasizes the need for transparent reasons; and Al Buhaira demonstrates that contextual legal interpretation cannot simply be reduced to textual computation. (DIFC Courts)

Therefore, the UAE's developing model is better described not as “replacing lawyers and judges with computers,” but as a movement from purely narrative legal analysis toward data-assisted, computationally structured legal reasoning.

The fundamental hierarchy remains:

Data → Computation → Evidence → Human evaluation → Legal rule → Reasoned judgment.

The computer can increasingly perform the calculation; the legal system must still justify the judgment.