The December AUKUS ^[1] Defence Ministers meeting in San Francisco has reinforced the importance of advanced undersea warfare capabilities as a key element of the agreement’s Pillar 2. A particular focus was the role of autonomous systems at sea—on and under the waves—together with AI in responding to future undersea threats.

A joint statement ^[2] emphasised maritime autonomy and experimentation through a series of exercises to ‘…enhance capability development, interoperability, and [increase] the sophistication and scale of autonomous systems in the maritime domain.’ These exercises would ‘refine the ability to jointly operate uncrewed maritime systems, share and process maritime data from all three nations, and provide real-time maritime domain awareness to support decision-making.’ It also talked about demonstrating and deploying ‘…common advanced AI algorithms on multiple systems, including P-8A maritime patrol aircraft ^[3] , to process data…and allow for timely high-volume data analysis.’

There was mention of UUV undersea launch and recovery, and quantum technologies to complement space-based positioning, navigation and timing services at sea. The role of AI in particular was prominent with a focus on ‘enhancing forcing protection, precision targeting, and intelligence, surveillance and reconnaissance’ across land and sea.

If these steps are pursued in full they could dramatically change how Australia approaches undersea warfare, centered on its planned nuclear-powered but conventionally armed submarines (SSNs). It’s important to emphasise that

The Navy will acquire three to five US Viriginia class SSNs from 2033 onwards and more of the SSN AUKUS in the 2040s. They will not by themselves be sufficient for Australia’s undersea warfare, or to deliver the ‘impactful projection’ Defence Minister Richard Marles wants. Only eight SSNs are to be in service. The ‘three to one’ rule allows for ^[4] two to three boats being available for operations at any time. It would be a mistake for Australia to base a notion of ‘impactful projection’ on just one platform, the SSN, or to reorganise its entire navy around an assumption that the SSNs will be the ‘war winning’ capability that can enable an effective ‘deterrence by denial’ strategy alone. The future navy needs greater combat mass and firepower if it is to contribute effectively to such a strategy.

With elements of the existing fleet aging or undermanned ^[5], there’s an urgent need to move more rapidly to build up Navy’s capability in the face of a rapidly deteriorating strategic outlook. The Navy’s surface combatant review is meant to fix this, but it won’t be released publicly until early 2024. Public comment on the review could feed into the National Defence Strategy ^[6] which is due to be completed by mid  2024. If the review does not significantly expand the Navy’s size and firepower, it  will be a missed opportunity in the face of rapidly increasing threats.

There needs to be a dramatic acceleration in development of advanced autonomous systems at sea,  both on and under the waves, with greater emphasis on smart and intelligent capabilities that fully employ AI, leaving humans strictly ‘on the loop’ in an oversight and managerial role, rather than directly controlling a platform remotely. The ‘tooth to tail’ ratio in terms of workforce to capability and effect needs to be reversed so that human oversight does not require large numbers of people to manage a few systems. The goal should be small teams managing significant numbers of uncrewed underwater and surface craft.

The AUKUS experimentation on AI in autonomous systems is intended to enable AI to take much of the load off humans. It will process information flowing from sensors aboard autonomous platforms on the ocean, such as Australia’s Ocius Bluebottle Uncrewed Surface Vessel, or under it with Anduril’s Ghost Shark.

Ultimately, AI needs to developed to the point where it can be given a greater role in managing the day-to-day operation of autonomous systems, with human oversight remaining essential for major decisions such as the use of lethal force. AI can contribute at the tactical level, as noted in the statement, on platforms such as P-8s or warships, but will have key roles to play in helping commanders and governments interpret complex data in a fast-moving operational situation potentially over vast areas. The goal should be rapidly gaining a knowledge edge, denying that same edge to the adversary, and acting within a decision-cycle of ‘observe, orient, decide and act’ faster than the opponent across a full multi-domain operational environment.

AI can do this far faster, at machine speed, than its human counterparts but human oversight and authority will still be needed at command and political leadership levels.

These advancements in AI will take us to tomorrow’s navy, but without expanding the size and firepower of naval capabilities, crewed or autonomous, they will not contribute sufficiently to enhancing Australia’s maritime security interests. A small navy with 12 surface combatants and eight submarines is insufficient to meet the challenges ahead. China’s PLA Navy is now the world’s largest, and is rapidly closing qualitative gaps on the US Navy.

Defence needs to be bold and ambitious and recognise that autonomous systems give us the ability to significantly boost fleet size, and potentially enhance our ability to bring firepower to bear at long range, if we are prepared to consider armed autonomous systems. Government needs to fund such an ambitious new navy or risk Australia’s security in the coming decade.

The central role of AUKUS allows all three partner states to investigate the full range of possibilities for AI together with autonomous systems, including armed autonomous USVs and UUVs. Establishing a network of autonomous and crewed systems that operate as a team across a maritime battlespace with an ability to detect, track and kill a threat on the surface or underwater, has to be the goal. Emphasising that combination of AI and autonomous systems working in concert with crewed platforms—and with critical human oversight ‘on the loop’—is the logical path to meet a potential challenge of a much more capable and assertive adversary with ambitious plans across the Indo-pacific, and with a potential ability to interfere with Australia’s critical maritime trade.