First Principles & Systems View: The Essential Twin Lenses for AI

Over the years, I've thought long and hard about what AI is.

Here we have the ability to replicate or outsource aspects of our own intelligence in a non-biological sense. That's what fascinates people most about AI - regardless of their background - and that's why it's a topic that has the ability to move or affect us all.

Something Karl Popper said resonated with me many moons ago:

"I think that there is only one way to science - or to philosophy, for that matter: to meet a problem, to see its beauty and fall in love with it; to get married to it and to live with it happily, till death do ye part - unless you should meet another and even more fascinating problem or unless, indeed, you should obtain a solution. But even if you do obtain a solution, you may then discover, to your delight, the existence of a whole family of enchanting, though perhaps difficult, problem children, for whose welfare you may work, with a purpose, to the end of your days."

AI is the problem that I've fallen in love with. And the more I uncover it, the more intriguing and fascinating it gets. That said, I'm consistently trying to sharpen the tools through which I analyze this particular technology and the domino effects it might have on society.

Two philosophies have consistently stood out in my journey with AI: the First Principles perspective and the Systems View.

A Dive into First Principles:

First Principles thinking is not new. Rooted in ancient Greek philosophy, it was notably championed by philosophers like Aristotle. They believed that to truly understand and resolve a problem, one had to break it down to its most basic elements. This philosophy has been adopted by various thinkers throughout history and has found its modern champion in entrepreneurs like Elon Musk.

In the context of AI, First Principles pushes us to declutter, to ask foundational questions like:

• What is intelligence?
• How do humans learn, and how can machines replicate or complement this?
• What constraints or limits decision-making (be it machine-driven or human-driven)?
• What is AIs fundamental nature and purpose?

We can take this a step further and apply it to a particular problem - let's take for example the Bias in AI, especially in areas like hiring/recruitment.

Understanding through First Principles:

• What is bias? In AI, it often emerges from historical data.
• How does AI learn? It studies patterns, so if data contains biases, the AI amplifies those biases.
• What are the objectives of hiring tools? To identify best-fit candidates based on skills, experience, and potential, not irrelevant factors.

Why is this so crucial? As AI systems become more integrated into our daily lives, understanding their foundational elements becomes non-negotiable. It ensures our solutions aren't merely iterative but are innovative, breaking the mold to deliver unparalleled value. But more importantly, the foundational elements can help us develop a matrix to understand what types of AIs we can apply within an organization, and for what kind of decision-making. This will help clarify the process of AI integration within a particular organization or entity.

A Dive into the Systems View of Life

If first principles give us clarity on the pieces, the systems view enables us to understand their interplay. AI doesn't exist in isolation. It's part of a complex web involving human interaction, societal norms, economic drivers, and much more. I was first introduced to the Systems View through the work of academic and philosopher Fritjof Capra.

Taking a systems view means recognizing the butterfly effect inherent in AI solutions. A change in one part of the system can have cascading impacts, sometimes in areas we hadn't even considered. This perspective ensures that as we innovate, we remain cognizant of the broader repercussions of our actions, striving for solutions that are not only effective but also ethical and sustainable.

Let's go back to our example of understanding the problem of Bias in AI from a Systems View.

• Historical Data & Societal Structures: This data is shaped by biased historical societal decisions.
• Feedback Loops: If a tool is biased and is used, the tool's future updates will further embed the bias.
• Economic & Social Implications: Overreliance can skew workforce demographics and perpetuate socio-economic disparities.
• Reputation and Legal Repercussions: Using biased AI tools may lead to legal challenges and missed opportunities for diverse talent.

Thinking about AI requires both depth and breadth. While first principles allow us to delve deep into core problems, a systems view ensures we don't lose sight of the broader picture, considering the unintended consequences and impacts of AI implementations. In a rapidly evolving world of AI, these twin lenses offer a compass, guiding us towards solutions that stand the test of time and mitigating the potential pitfalls of a myopic view.