Leaders need AI but does AI need Leaders - REAIM23 Perspective

Thoughts relating to REAIM 23 Presentation - Leaders need AI but does AI Need Leaders at The Hague.

When we talk about AI, we keep talking about NOW and the impact on ME as an individual. It is a very personal, immediate reference point and often falls into a binary test: is AI affecting me now, and if not, then it is nothing to worry about.

We also seem to have a singular view about war always being the same and universal, a bloody, violent mess resolved by courageous individuals grappling in mud. If technology is to change war, then it will only make it faster, more brutal, and more violent.

At REAIM23, we have heard historians saying that warfare was fixed by how we learned to talk 100,000 years ago and that writing 6,000 years ago compounded the problem. We have heard military leaders proclaim that the values which enabled them to progress in their careers are the same that we will need in the future.

We seem determined to convince ourselves that warfare will not change.

Since 2015, there has been a series of predictions that would show AI is fundamentally changing the world. Taken from several documents, as only some things can be summarised with a few bullets, I have compiled the key steps below. Our world continues to progress through those predictions, for good or bad:

• Demonstrate that AI can achieve better than human skills like translate and image recognition
• Build the infrastructure that would enable global scale and capacity with cloud storage and compute
• Construct component building blocks that would enable the adoption of AI across sectors and industries
• Democratise AI by making it easy to implement and access through packaged capabilities to generate content and output
• Interface with AI using conversational and intuitive models that empower anyone with access
• Replace mundane repeatable activities and tasks to enhance human ingenuity
• Agree that AI is changing society and that it needs international collaboration on its introduction and control
• Test AI to show that it acts in unpredictable and unintended natures
• Create common principles and approaches to develop safe and trustworthy AI
• Transition human activity in key sectors with AI alternatives that reduce costs and increase AI development
• Adopt AI in high-risk areas like security, justice, and defence to improve performance and reduce (own side) military casualties
• Use AI to develop and scale future AI performance and adoption within and across roles, functions, and sectors
• Support humanity as they transition from work that involves mundane, repeatable activities into more creative, insightful activities
• Increase digital skills to anticipate and adapt to working alongside AI toolsets.
• Develop international and national planning, funding and support for people who are no longer employed or employable
• Anticipate highly automatable sectors to help those affected transition to employment elsewhere
• Encourage that the right mindset about AI is more important for a safe long-term transition than understanding only the technical toolset
• Plan for a society that enhances human ingenuity with AI that empowers human life with value and worth

In all these cases, we have taken the easy path, taking the parts that reduce costs or deliver immediate gains, ignoring the more complex elements like international agreement, and are yet to consider the consequences in a meaningful, planned, and funded way. REAIM in 2023 is perhaps a key step but also one that seems late.

We've stripped out the easy, taken the quick gains, and left future generations to pick up the bill.

It becomes increasingly hard to examine this list and argue that we are not changing society. Due to that revolution, we must change vast elements of warfare and the military. This change should come from within militaries that capture and use the benefits of AI but will probably come from outside militaries either from their new recruits or, more likely, by their adversaries.

As an evangelist, I feel that if you are a military leader today and you are NOT preparing for an automated future then you are failing to prepare your command and those who will follow you. This is more than accepting the perceived wisdom that AI will not change how warfare is conducted or that warfare is universally the same.

It requires a deeper consideration of how your command could be changed and acting on those possibilities.

At times, it feels that sharing this warning is more like Niels Bohr publishing about quantum atomic theory, as it is hard for society to visualise the inevitable consequences. Yet the world had thirty years before Oppenheimer took those theories and began discussing destroying worlds with atomic bombs.

Today, unlike the atomic age, the research time for disruptive AI from theory to deployment is measured in months and not decades

There are three factors that leaders should consider when thinking about whether AI needs leaders:

We need AI to compete and win

We cannot process the volume of data, deliver valuable insights, or operate at the tempo we need to succeed without AI support for our military services. This is an absolute based on similar experiences in other sectors. Yet, current approaches are planning to add a bit of AI here and a bit there without truly considered thought or detailed assessment. I have written elsewhere about the penny-packet approach to AI. I would add that industry doesn't always help by encouraging users to look at fantastic new tools to buy.

We need AI to win, and we need to implement that AI across whole swathes of military operations.

Second, AI changes every sector and industry with flatter hierarchies and simpler organisational structures.

AI democratises decision-making, accelerates sharing, and increases the capacity of individuals and teams. Teams are more productive and operate faster supported by AI and therefore do not need the same level of management supervision and review.

In the same way that robots replaced manufacturing lines staffed with workers, supervisors, inspectors, line managers, production managers and the myriad of middle management, AI will do the same for organisations that use and exploit information, data, and insights.

The biggest threat from AI is targeted at middle managers.

Finally, anticipate that future leaders will have very different development paths from our current leaders.

Military leaders continue to see the path to future promotion as the path that they follow. Advice is passed down to learn in this staff appointment, develop in this command position, and operate in this environment.

One day, promotion will be yours, as long as you follow the way

Yet we see that automation is changing, replacing, rebuilding, and restricting those paths. We cannot plot a path for future leaders, who will face very different structures, challenges, and ways of resolving complexities, with a way that ignores these critical changes. We need to consider now what that path looks like and make the changes today.

 Unless you're in an organisation that is riddled with hubris, changes slowly, adopts technology to accelerate what they do today rather than change how they do it, defers to hierarchial command and control, and plans for change over decades rather than months.

Still, for the leaders who remain in those organisations, where do they come from, and how do they develop? Successful organisations must plan, select, train, and change to offer future leaders valuable experiences that test and develop their leadership skills, that value their human ingenuity, and still reward their endeavours.

At the same time as automating and replacing many of those experiences.

The key conclusion from asking does AI need leaders is that in an automated world, where we have replaced the easy and mundane tasks but still require leadership to face the most complex of challenges, how and where do our future leaders learn to meet those demands?

Leadership is about preparing your teams for the future. Preparing for the future is also where AI needs leadership.