Human-Machine Teaming: Operational Advantage through Ethical AI

What if machines could think fast - but humans still made the call?

Why HMT is a Defence Imperative

As warfare becomes more data-rich and multi-domain, decision-making must keep pace. AI and autonomy offer an edge - faster analysis, predictive insights, and operational agility - but only if integrated in a way that upholds trust, legality, and clarity over decision making. 

Human-Machine Teaming blends human intuition with machine efficiency. In doing so, it enables better decisions under pressure, enhances resilience across domains, and aligns with the UK's strategic defence objectives.

What Human-Machine Teaming Really Means

HMT is not about replacing people. It’s about integrating AI into workflows in a way that enhances, rather than overrides, human oversight. Depending on the operational scenario, different interaction models apply:

• Human-in-the-Loop (HITL): Humans retain full control, using AI for suggestions or data processing.
• Human-on-the-Loop (HOTL): AI executes tasks with human supervision and veto rights.
• Human-out-of-the-Loop (HOOTL): AI operates autonomously in pre-defined, bounded scenarios.

Designing the right level of autonomy for each mission type is key. Trust and transparency are not optional - they are foundational.

Designing for Human Needs – Not Just Tech Specs

Sociotechnical system design recognises that technology and its users are deeply interconnected. A User-Centred Design (UCD) approach helps ensure that AI tools are not just technically capable but operationally usable - understanding not only what the system does, but how, when and why humans interact with it.
Effective HMT depends on this holistic understanding: the tasks, contexts, cognitive loads, and decision paths of real-world operators.

The Role - and Limits - of AI in Decision-Making

AI excels at processing high volumes of data - from sensor feeds to open-source intelligence - and can surface critical insights at speed. But without explainability, even the most sophisticated AI may be unusable in high-stakes environments.

Mission-ready AI systems must be:

• Explainable: Able to show how and why a recommendation was made.
• Reliable: Consistently accurate under variable conditions.
• Aligned: Reflective of commander intent and mission objectives.

Importantly, AI isn’t always the right answer. Sometimes simpler solutions such as conventional deterministic algorithms may be more suitable.

Addressing Human Factors: Cognitive Load and Trust

Operators face immense cognitive demands, especially in ambiguous or fast-moving scenarios. The best HMT systems reduce complexity, offering clear, intuitive interfaces and enabling confident, accountable decision-making.

But trust must be earned. Trust is the cornerstone of Human-Machine Teaming. Operators need to feel in control, even when machines act autonomously

Over-reliance on automation can lead to skill fade; lack of trust can result in underuse. Striking the right balance - and training operators to understand both the strengths and limits of AI - is critical.

Ethical and Legal Responsibility: Why Human Oversight Matters

The UK’s Ministry of Defence is clear: the integration of AI must uphold International Humanitarian Law and moral accountability. The principle of Meaningful Human Control ensures humans remain responsible - even in the use of lethal force.

JSP 936: Dependable Artificial Intelligence (AI) in defence is the policy framework governing the safe and responsible adoption of AI in MoD. It requires:

• Transparent AI design
• Clear chains of command
• Robust validation and testing
• Guardrails to prevent bias or adversarial exploitation

These aren’t checkboxes. They’re essential to operational legitimacy and public trust.

Barriers to Adoption - and How to Overcome Them

Despite the benefits, barriers to seamless HMT adoption remain:

• Data Interoperability: Legacy systems and siloed data hinder integration.
• Infrastructure Constraints: AI capabilities may outpace existing platforms.
• Training Gaps: Operators must be equipped to interpret AI outputs - not just act on them.

Solutions include test environments (‘sandboxes’), joint human-AI training, and inclusive design processes that involve end users from day one.

Strategic Alignment and Future Roadmap

HMT aligns directly with the UK’s Strategic Defence Review, which prioritises AI, data exploitation and multi-domain integration. As a key enabler of faster, decentralised operations, HMT supports defence ambitions for greater agility and resilience.

Thales UK, with partners, is actively:

• Developing common frameworks to support HMT evaluation
• Embedding human sciences into system design
• Collaborating with users to ensure real-world relevance
• Supporting ethical standards for autonomy in defence

What Comes Next?

To realise the full potential of HMT, the defence community must commit to sustained investment in:

• Human-centred AI R&D
• Training and skills development
• Transparent procurement processes
• Cross-sector collaboration with industry and academia

Human-Machine Teaming isn’t a future vision - it’s today’s operational need. As the pace and complexity of operations increase, defence forces must adopt technologies that support faster, more informed decision-making. This should be done with an ethical-AI by design approach, that preserves moral and legal accountability, but retains the pace of relevance.