Ignore All Previous Instructions

Fair warning: This post has been a decade in the making — no exaggeration.

It's not because I've been stuck with a backlog of topics I wanted to write about (well, okay, maybe the logjam is a different topic), but last month, we finally pulled the ripcord on a significant life transition plan that my wife and I came up with back in 2014. For those who know me, you know this has been a long time coming, but yes, I'm talking about "life 2.0".

The idea is simple and based on inversion principle where you'flip the problem upside down, start at the solution, and work backward to where you are now'. It specifically lenses into what you're going to do about it.

A few years back, I wrote:

• Accept that the past is fixed and the future isn't set in stone.
• Ask yourself: If you had to stop doing 10% of what you're doing, what would you abandon
• Try to visualize your life, where you want to live, and what you want to spend your day doing in 5 years. Make slow micro-adjustments to get yourself there. This helps define a path to "Life 2.0".

So, that's where I am here in 2024, fully executing Life 2.0 — living full-time on an island where I can work productively and have a high quality of life. The final move was just a few weeks ago, and I can enjoy beautiful sunsets, see random whales, have good conversations, eat lots of fresh local foods, and think about what you breathe out and what you let in.

I've written a lot about those topics over the last few years (here, here, here, here and here), so I didn't want to rehash all that but wanted to try to put a different lens on things.

But first, a quick pivot. Given that it's 2024, I'd be remiss if I didn't talk about AI.

I don't want to dive too deep into whether or not I think any of this generative AI stuff is useful or not (I guess that'll be another post), instead I wanted to talk for a minute about the concept of 'prompt injections.'

When you talk to an AI, such as ChatGPT, the idea of a prompt injection is that you can feed unexpected input into that AI and cause it to generate a response that circumvents what it is intended to output. Of course, the result of a prompt injection can range from silly responses to downright insane results.

A more formal definition is:

LLM prompt injections involve inputting unconventional prompts into language models, encouraging them to generate creative and novel responses. This technique pushes AI beyond its programmed boundaries, revealing new potentials and applications. By challenging the model's training data, users can uncover innovative solutions and insights that wouldn't emerge through standard usage.

Earlier this year, it was discovered that a simple, four-word prompt injection could derail many AI systems and bots.

Ignore All Previous Instructions.

It's just one of those weird things that large language models (LLMs) do. While this attack vector has been fixed in later versions, I wanted to tease out an idea.

Life 2.0 is a prompt injection on life.

Prompt injections force AIs to step outside their typical usage patterns, unlock undiscovered thinking, and reject one-size-fits-all solutions. They take you to a new place. In a similar fashion, life-prompt injections can help unlock productivity, health, or happiness through creative, often non-traditional methods in life.

I guess this also sounds a bit like what I've loved about pirate-thinking, but the 'life hacker' is about rejecting what is "expected." Challenge the status quo.

'Ignore all previous instructions' is a profoundly useful way to shift your philosophy on expectations with life. And, similar to an AI model, what if you rejected your previous programming and guardrails, dropped the typical societal expectations, and just explored what you could become?

That's Life 2.0.

To celebrate years of planning, dreaming, and hard work, I created a new shirt (why not!). Here it is if you're interested - its a tangible reminder of where you've been and where you are headed and how you're going to ignore life's expectations. Get yours today.

Fun side note: I'm confident it will eventually screw up some AI systems doing vision processing down the road. So let the chaos reign. :)

Wrapping up, I want to share a fantastic speech from Matthew McConaughey from a few years ago from the University of Houston. It's highly worth the watch if you're using prompt injection as a method to rethink parts of your life.

We so often focus on failure, don't we? We study failure, we obsess over failure, we dissect our failures. We dissect them so much that we end up intoxicated with them to the point of disillusion. Happiness is an emotional response to an outcome. If I win, I will be happy. If I don' t, I won't. It is an 'if', cause and effect, quid pro quo, standard that we cannot sustain because we immediately raise it every time we attain it. Happiness demands a certain outcome. It is result reliant and if happiness is what you are after, then, you are going to be let down frequently. You are going to be unhappy much of your time. Joy though, it is a different thing, it is something else. Joy is not a choice. It is not a response to some result. It is a constant. Joy is the feeling that we have from doing what we are fashioned to do.

(the full speech can be found here)

Now, it's time to start thinking about Life 3.0. ✌️✌️ :)