Niraj Chaurasiya

Building products, research, and frameworks to explore how intelligent systems think, learn, reason, imagine, evaluate, and build under uncertainty.

[Niraj Chaurasiya @Nirajbuilds]

Today I realized every project I’ve built over the last year was secretly trying to answer the same question.

It took rebuilding my entire website to see it.

🧵

[Niraj Chaurasiya @Nirajbuilds]

Interesting @elonmusk

[Elon Musk @elonmusk]

Turns out the future does have flying cars

[Elon Musk @elonmusk]

Turns out the future does have flying cars

[Niraj Chaurasiya @Nirajbuilds]

Have you ever started reading a book with excitement, set a deadline, and then lost motivation halfway through?

[Niraj Chaurasiya @Nirajbuilds]

Can I Explain This? | Ep. 01
Identity-Based vs. Outcome-Based Habits

[Niraj Chaurasiya @Nirajbuilds]

A model is coherence in the mind.
The next step is to test it scientifically.

Building Systems Under Uncertainty

[Techshortsapp @techshortsapp]

Next step: scientifically verifying TechShortsApp.

Moving from a conceptual model to empirical validation—testing how much evidence behavioral signals actually provide about video credibility against expert-rated ground truth.
@techshortsapp

[Niraj Chaurasiya @Nirajbuilds]

Who will remain the author of your life?

Project Agency — The Beginning

#ProjectAgency #HumanAgency #ArtificialIntelligence #HumanCenteredAI #DecisionMaking #AIResearch #FutureOfAI

[Niraj Chaurasiya @Nirajbuilds]

Collection of better distinctions:

observation vs evidence,
knowing vs learning,
mechanism vs measurement,
state vs representation,
function vs experience,
intuition vs empirical weight.
My ignorance didn’t disappear. It acquired structure.

— Building Systems Under Uncertainty

[Niraj Chaurasiya @Nirajbuilds]

I came searching for better AI.

Somewhere along the way, I realized I was searching for the assumptions behind intelligence itself.

[Niraj Chaurasiya @Nirajbuilds]

The crawler collects evidence.
The model builds representations.
The system estimates confidence.
Intelligence begins long before prediction.

[Niraj Chaurasiya @Nirajbuilds]

Experience updates memory.
Evidence updates belief.
Together, they shape how an intelligent system understands reality.

[Niraj Chaurasiya @Nirajbuilds]

Sometimes I wonder if I'm studying artificial intelligence.

Other times I realize I'm studying the principles that make any intelligence possible.

[Niraj Chaurasiya @Nirajbuilds]

Progress isn't replacing questions with answers.

It's replacing vague questions with precise ones.

[Niraj Chaurasiya @Nirajbuilds]

I used to think interdisciplinary work meant knowing many disciplines.

I’m starting to see it differently:Questions are the driver. Domains are tools.

Use whatever disciplines the problem actually demands—and go deep enough to use them responsibly.

[Niraj Chaurasiya @Nirajbuilds]

I'm beginning to think my research isn't about educational videos.

It's about how intelligent systems construct reliable beliefs from uncertain evidence.
@techshortsapp

[Niraj Chaurasiya @Nirajbuilds]

We don't directly observe intelligence.
We don't directly observe learning.
We don't directly observe credibility.
We infer them.
The real question is:

How should an intelligent system infer latent variables from uncertain evidence?

[Niraj Chaurasiya @Nirajbuilds]

Google doesn't know Isaac Newton.

It infers an internal representation of Newton from accumulated evidence.

Humans may not be as different as we think.

[Niraj Chaurasiya @Nirajbuilds]

The most dangerous assumptions are the ones we never notice.

The most valuable questions are the ones that expose them.

[Niraj Chaurasiya @Nirajbuilds]

Science begins when assumptions stop being invisible.

[Niraj Chaurasiya @Nirajbuilds]

Chalmers made one distinction difficult to unsee:

Explaining what a system can discriminate, report, learn, remember, or control does not automatically explain why any of that is accompanied by subjective experience.

Function is not obviously experience.

[Niraj Chaurasiya @Nirajbuilds]

Before asking what an internal representation is, ask:

Why should we assume it exists at all?

Every hidden variable deserves evidence before it deserves a theory.