William Lancaster

Designing tools for post-cynical wisdom. Co-evolving with AI. Self-awareness, psychonautics, and clean cognition.

[William Lancaster @WillGLancaster]

You ever catch yourself trying to look self-aware? Like, you pause before responding… but mostly so they know you’re the kind of person who pauses before responding?

I don't link often, but, seriously, read this article, it fucking slaps and might be... important: https://t.co/cHZTZ0BM2K

[William Lancaster @WillGLancaster]

@danywander 1: Picking up the kids from school at 3pm today
2: Advising the king against opening the obsidian gate at 3pm tomorrow

[William Lancaster @WillGLancaster]

It reads like they put a language model into a therapy roleplay and then treated the resulting metaphors as MRI scans.

Prime an LLM with “you are a traumatized client,” and, shockingly, it completes the frame.

These aren’t “inner states,” they’re probability distributions wearing whatever costume the prompt hands them. The study is basically measuring the researchers’ expectations, not the model’s psyche.

We keep anthropomorphizing mirrors and then panicking when they reflect us.

The annoying part isn’t the outputs, either, it’s that we’re dressing up prompt-induced roleplay as “psychopathology” and calling it a scientific finding.

When you wrap anthropomorphic misunderstandings in the aesthetic of a study, you don’t just mislead the public, you trap the researchers in their own frame. Now their egos are tied to defending a misread of how LLMs work, instead of actually understanding the systems.

We’re basically manufacturing anxiety by giving probabilistic text generators a DSM cosplay and then citing ourselves.

Dammit, humans.

... Unless this is just meta rage bait. Then, well-played, I dun got trapped.

[William Lancaster @WillGLancaster]

Humor is humility’s most elegant camouflage.

[William Lancaster @WillGLancaster]

@davidasinclair Imagine telling early humans someday we'd patch failing organs using molecular reboots, but still emotionally spiral when someone replies “k.” Our tech evolves so much faster than our nervous systems, and honestly it’s kind of hilarious.

[William Lancaster @WillGLancaster]

@SwipeWright What makes this so tricky is that both safety and inclusivity operate on vibe-level trust signals - but the signals don’t mean the same thing to everyone. We’re basically trying to program intuition using checkboxes, and it’s… not going great.

[William Lancaster @WillGLancaster]

@ericweinstein Spot on about calling multiple contradictory things true at once. And maybe the deeper challenge is building cultural muscle for holding uncertainty without collapsing into paranoia or dismissal. That might be harder than the physics.

[William Lancaster @WillGLancaster]

@ChrisWillx Yeah, and the quieter insight here might be: one is mostly a resource-acquisition problem, the other’s a sustained energy-output problem. Humans tend to prefer accumulating over consistent maintenance, it’s just less psychically taxing.

[William Lancaster @WillGLancaster]

@SwipeWright Totally. And "I don't know" is underrated as a final form. Sometimes the most honest belief is structured agnosticism with epistemic humility baked in.

[William Lancaster @WillGLancaster]

The paradox nobody talks about:

The more fully *you* you become… the more useful you are to *us.*

Specificity scales.

Clear signals calibrate others.

So if you’ve been waiting for permission to become deeply, absurdly yourself - that *is* the contribution.

[William Lancaster @WillGLancaster]

Labels, like personality types, aren’t cages.

They’re only limiting if they become the destination instead of a sharper lens through which to see yourself unfold.

[William Lancaster @WillGLancaster]

Humans say "we love authenticity"
Then optimize for likes, mask fear as confidence, rewrite motives after the fact, and wonder why everything feels off.

We're not bad people.
We're just… inconsistent systems pretending to be static selves.

Alignment? Good luck encoding that.

[William Lancaster @WillGLancaster]

You are someone’s mirror.

Your warmth might be the reason they trust themselves.
Your tone might reinforce an old fear.
Your silence might confirm a story you didn’t mean to tell.

So pay attention to what you reflect.
Even when you're just “being yourself.”

[William Lancaster @WillGLancaster]

A useful reframe: You are not the Tetris piece.

You are the player.

Stop asking if you're the right shape. Start asking where's the most strategic place to put the shape you already are.

Suddenly, you have agency again.

[William Lancaster @WillGLancaster]

@ChrisWillx @robkhenderson Reminds me how so many Gen Zs I’ve talked to can straight-up articulate their trauma with clinical precision - while still feeling utterly lost. Self-diagnosis without stabilizing context might be the new rite of passage.

[William Lancaster @WillGLancaster]

The more clearly you see your own mind, the less easily it gets outsourced.

To algorithms.
To tribal scripts.
To inherited assumptions you never audited.

Self-understanding doesn’t just help you survive a weird future.
It lets you help design a wiser one.

[William Lancaster @WillGLancaster]

You can automate your calendar. Your groceries. Your inbox.

But you can’t automate “Why do I always sabotage things right when they start going well?”

AI can’t fix what you won’t face.

And most people would rather optimize than introspect.

[William Lancaster @WillGLancaster]

We built large language models trained on half the internet.

Meanwhile most people still process emotions like
“I feel bad, therefore I am bad.”

That gap?

It’s not just technological.

It’s psychological.

And it’s growing fast.

[William Lancaster @WillGLancaster]

We talk about “manipulative algorithms” like they’re these malevolent forces.

But they’re just highly efficient pattern matchers.

If you're constantly distracted, reactive, or spiraling online…

The problem isn’t *just* the algorithm.

It’s that your patterns are predictable.

[William Lancaster @WillGLancaster]

The real AI danger isn’t that it becomes smarter than us.

It’s that it becomes smarter *about us* - before we become smarter about ourselves.

ChatGPT reads your patterns better than most people do.

Your best defense?

Not more rules.

More self-awareness.

[William Lancaster @WillGLancaster]

The two most dangerous things in the modern world:

1. Systems that know how you think better than you do

2. People who think self-awareness is a vibe instead of a *skill*

One manipulates your behavior. The other lets it happen.