Research of the Day HUMBL Voice · Intelligence File 006
Research of the Day

August 2026

The OpenAI India Test

Sixteen questions. Not one asks for a resume.

Intelligence FileOpenAI IndiaCreative HiringJob Applications

OpenAI India put up a creative role. The application is sixteen questions long, and it is really a test of judgement in disguise. We took it apart, question by question, and mapped what each one is actually measuring.

16 Questions
0 Ask for a resume
6 Banned words
12 Word cap, hardest one
Community bonus

Got the code from the video?

Enter it with your email to unlock the deeper cut, keep a PDF of this file, and claim founding-member access on the HUMBL app.

Enter your code + email to unlock this file's bonus.

Not from Instagram? The code's hiding in today's video, not here.

Watch it on Instagram
01

Every question is a cage.

This was not a job form. It was a sixteen part test of creative judgement. A portfolio tells you what someone made. This form tells you what they will make when the brief is bad, the deadline is tomorrow, the obvious words are banned, and they still have to produce something worth noticing.

A portfolio tells you what someone made. This form tells you what they make under constraint. Every question is a cage, and the cage is the point.

02

Part one — can you make things (Q1–Q6)

Q1

“Tell us about one piece of work you are genuinely proud of. Share a link. What exactly was yours, and what is the one decision you are proudest of?”

What it measures

Ownership. Not can you point at good work, but can you isolate your specific contribution and the single decision behind it. Vague "we" answers fail here.

Q2

“Now show us something you made that bombed. What did you learn? What would you change if you made it today?”

What it measures

Whether you can look at failure without flinching or spinning. The tell is whether the "learning" is real or a disguised brag.

Q3

“What did your last team always come to you for? What would they absolutely never come to you for?”

What it measures

Honest self knowledge. The first half is easy ego. The second half forces you to name a real edge you do not have, out loud, without humblebragging. Most people cannot.

Q4

“What is one thing on the internet recently that made you jealous you did not make it?”

What it measures

Your taste, and your reference set. What you envy reveals the ceiling you are aiming at better than anything you claim to admire.

Q5

“What is something you have gone down a completely unnecessary rabbit hole learning?”

What it measures

Genuine curiosity versus performed curiosity. "Unnecessary" is the key word. They want obsession with no payoff, because that is what real interest looks like.

Q6

“Tell us about one piece of feedback you ignored. Were you right?”

What it measures

Conviction and calibration. Can you hold a line against feedback, and can you be honest about when holding it was a mistake.

03

Part two — can you make this thing (Q7–Q12)

Q7

“Share a link to the most interesting conversation you have had with ChatGPT.”

What it measures

Whether you actually use the product, and how. The conversation you pick is a portfolio of how your mind works with the tool.

Q8

“Make one deeply boring thing famous by tomorrow. What is it, and what is the first post?”

What it measures

The core creative skill for this job. Can you find the angle in something with no obvious angle, and can you start, not just strategize.

Q9

“Explain ChatGPT to your mother, your ex, and a 14 year old. One sentence each.”

What it measures

Audience range. Three registers, three relationships, three levels of context. It is a compression test wearing a joke.

Q10

“Make someone want to try ChatGPT without using these words: AI, future, smarter, revolution, unlock, possibilities.”

What it measures

Whether you can sell without the crutch words. Banning the obvious vocabulary is how they find out if you have real vocabulary underneath.

Q11

“A creator comments: "This sounds like ChatGPT wrote it." Reply from the brand account. Twelve words maximum.”

What it measures

Grace under hostility, in the brand's voice, under a hard cap. The twelve word limit is the real test. No room to hide behind length.

Q12

“Pick one Indian creator to run ChatGPT India for a day. What do they post first?”

What it measures

Cultural fluency and casting instinct. Do you know the room, and can you match a voice to the brand without breaking either.

04

Part three — do you have a point of view (Q13–Q16)

Q13

“What do you currently do, or what did you do most recently?”

What it measures

The one straight question. Placed late on purpose, after they already know how you think. Context, not credential.

Q14

“Turn your latest rabbit hole into a five slide Instagram carousel. Only give us the slide headlines.”

What it measures

Structure and restraint. Headlines only means you cannot lean on visuals or body copy. Can you carry a narrative on five lines.

Q15

“Which is the bigger win: one million views or 100 people copying the behaviour? You only get one sentence to defend your answer.”

What it measures

Whether you understand reach versus influence, and whether you have a spine. No right answer. Only whether you can defend a real view without hedging.

Q16

“Anything else we should know before speaking?”

What it measures

What you choose to do with an open door. The answer, or the choice to leave it blank, is itself the signal.

05

The three hardest.

Three questions carry the piece, and they are the ones the original post quoted.

“Anyone can answer what their team relied on them for. The second half, what they would never come to you for, is a landmine. Say nothing and you look self unaware. Brag in reverse and you look worse. The only passing answer is a true weakness stated plainly. It is measuring whether you can be honest when honesty costs you something.”

— Q3 · the honesty trap

“A hostile comment, a brand account, and a hard cap. Too long and you sound defensive. Too clever and you sound smug. The constraint is the whole test: it strips away everywhere a weaker writer hides and leaves only judgement. Can you be warm, sharp, and done in one breath.”

— Q11 · the twelve word cage

“One million views or a hundred people copying the behaviour. There is no correct answer and they know it. What they are watching is whether you can pick a side and defend it in a single sentence without softening it into "it depends." A point of view, held.”

— Q15 · the spine test
The Thesis

A portfolio shows you the work. A good question shows you the person.

Sources & methodology

OpenAI India creative role application, questions reproduced as posted.

Analysis and "what it measures" interpretation is HUMBL Voice's own.

Questions are reproduced as posted by OpenAI India. The analysis and interpretation of what each one measures is HUMBL Voice's own reading, not an official statement from OpenAI. Not affiliated with or endorsed by OpenAI.

HUMBL Voice · HUMBL.Voice · August 2026

Follow @humbl.voice for the next file