216. FUTURE CONSTITUTIONAL GOVERNANCE OF AI-OPERATED UTILITIES

1. Meaning and Constitutional Context

AI-operated utilities are public or regulated utility systems in which artificial intelligence performs or assists functions such as electricity forecasting, grid balancing, outage detection, tariff administration, infrastructure maintenance, demand management and customer-service decisions. Future constitutional governance must ensure that technological automation remains subordinate to constitutional rights, legality, accountability and public interest.

In South Africa, the constitutional framework is particularly important because utilities such as electricity and municipal services perform functions closely connected with socio-economic rights and public administration. Section 33 guarantees lawful, reasonable and procedurally fair administrative action, while section 195 requires public administration to be accountable, transparent, impartial, fair and responsive.

2. Constitutional Principles Governing AI Utilities

Future AI-operated utilities should therefore operate according to several constitutional principles.

First, legality: An AI system cannot exercise public power merely because it has been technologically programmed to do so. There must be a valid legal source authorising the relevant decision.

Second, administrative justice: Where an AI-assisted decision adversely affects a person's rights or legitimate interests, the decision-making process must satisfy PAJA and constitutional administrative-justice requirements. PAJA gives effect to section 33 and promotes accountability, openness and transparency in public administration.

Third, transparency and explainability: Section 195 specifically requires timely, accessible and accurate information and promotes transparency. Consequently, a utility should not be able to defend a rights-affecting decision simply by saying that its algorithm is a “black box.”

Fourth, equality and non-discrimination: Automated systems used for service allocation, disconnection management, credit assessment or infrastructure prioritisation must not reproduce discriminatory patterns contained in historical data.

Fifth, accountability: Responsibility must remain attributable to identifiable public officials, utility boards, regulators or contractors. Delegating a decision to an algorithm should not eliminate constitutional responsibility.

3. Case Law

Case 1: Pharmaceutical Manufacturers Association of SA v President of the Republic of South Africa

Citation: 2000 (2) SA 674 (CC); [2000] ZACC 1

Facts: The case concerned the legality of governmental executive decision-making surrounding the registration of medicines.

Legal Issue: Whether the exercise of public power complied with constitutional legality requirements.

Judgment: The Constitutional Court established that the exercise of public power is subject to constitutional control.

Legal Principle / Ratio Decidendi: Public power must have a lawful basis and must satisfy constitutional standards of rationality.

Significance: This principle is directly relevant to AI utilities. An algorithm cannot become an independent source of governmental power; its operation must remain connected to a lawful empowering provision.

Case 2: AllPay Consolidated Investment Holdings v CEO of SASSA

Citation: 2014 (1) SA 604 (CC); [2013] ZACC 42

Facts: The case concerned the constitutionality of a government tender for the payment of social grants.

Legal Issue: Whether procedural irregularities in a constitutionally regulated procurement process could invalidate the decision.

Judgment: The Constitutional Court emphasised that constitutional and statutory requirements governing public procurement are legally binding and that fairness and lawfulness must be assessed independently of the eventual outcome.

Legal Principle / Ratio Decidendi: Public administration must follow legally prescribed procedures; transparency, fairness and accountability cannot be sacrificed merely because an outcome appears efficient.

Significance: The principle is highly relevant when utilities procure AI systems. Algorithmic procurement, training-data requirements, auditing and performance standards should form part of a legally compliant procurement framework.

Case 3: Waco Africa v Eskom SOC Ltd

Citation: [2022] ZAGPJHC 631

Facts: The dispute involved Eskom's procurement process.

Legal Issue: Whether Eskom's procurement framework had to comply with constitutional and administrative-law requirements.

Judgment: The Court confirmed the importance of section 217, PAJA and legally binding procurement requirements in public procurement.

Significance: The case provides a useful constitutional foundation for future procurement of AI-driven electricity infrastructure, smart-grid systems and automated utility platforms.

4. Future Governance Model

Future constitutional governance should require human oversight, algorithmic audits, cybersecurity controls, explainable decision-making, impact assessments, records of automated decisions, independent review mechanisms and effective remedies. High-risk decisions—particularly electricity disconnections, essential-service allocation, tariff-related decisions and emergency grid interventions—should retain meaningful human accountability.

5. Conclusion

The constitutional future of AI-operated utilities is not simply about technological efficiency. It concerns whether automated public power remains lawful, fair, transparent, accountable and rights-compatible. South African constitutional law already provides important foundations through legality, administrative justice, equality, public participation and accountable administration. Future legislation and regulation can build upon these principles by imposing specific requirements for algorithmic transparency, human oversight, auditability and judicial review. AI may automate utility operations, but constitutional responsibility must remain with legally accountable institutions and decision-makers.