A screenshot landed in the group chat — an AI told someone to walk to a car wash a hundred metres away. It was the right answer. The circle spent the next half hour showing why it might also have been the wrong one.
It started, as these things do, with a screenshot. Rajesh dropped one into the circle's WhatsApp group — a throwaway exchange with an AI assistant, the kind you screenshot for the laugh and move on.
The question was almost too simple to ask: the car wash is a hundred metres away — should I walk or drive? The machine answered exactly the way a sensible friend would.
The screenshot Rajesh shared — rebuilt here so it stays readable.
A perfect answer. Walk, obviously. And yet the circle — a group of people who think for a living about how machines reason — didn't laugh and scroll past. They leaned in. Because the more you look at that little exchange, the more it stops being about a car wash at all.
Abhiram was the first to turn it over, and he turned it the right way round:
I believe it has given the right answer. The car wash is 100 meters. It is not clear if you are giving the car to wash, or you want to go there to collect the car that was already given to wash. So clear prompting is needed for AI to be effective.
12:17Clear communication — a key leadership trait — applies here too.
12:29Context is the key, be it in leadership (being context-aware) or in this case providing a prompt.
12:37Aswathama was killed — and Dronacharya gave up. The truth was told, but the context was buried.
The Mahabharat was won because of a half-truth.
Context is everything.
12:41Right answer, good to walk to collect your car.
12:45Read it again. The AI wasn't wrong — it was under-briefed. "Should I walk or drive to the car wash" hides an entire missing clause: are you going to drop off a dirty car, or to collect a clean one? Vaskar landed it neatly — good to walk to collect your car. But flip the intent and the same advice falls apart: drop-off, and you'd walk over only to find you've nothing to hand across the counter. The answer was flawless for a question nobody had quite asked.
And that, Abhiram pointed out, is not really an AI problem. It's a communication problem wearing an AI costume. The model did the most reasonable thing it could with what it was given. The gap was upstream — in the prompt.
Here is where the circle did its favourite move: it took a small, technical observation and walked it up the stairs until it became a leadership one.
Vikas said it plainly — clear communication is a key leadership trait, and it applies here too. The thing we demand of a good prompt is the same thing we demand of a good brief, a good delegation, a good piece of feedback: enough context that the person on the other end can act on what you mean, not just on what you said. Tuhin closed the loop — context is the key, whether you're being context-aware as a leader or context-rich as a prompter. Same muscle. Different room.
The half-correct answer about the car wash is exactly the kind of thing a smart, willing team member hands you when you brief them in a hurry. They did precisely what you asked. The fault, if there is one, points back at the asker.
Then Vikas reached for the oldest example in the book — and it landed so hard the chat lit up.
In the Mahābhārata, the guru Droṇa was unstoppable in battle — until the Pāṇḍavas needed him to lay down his arms. The plan: tell him his son, Aśvatthāmā, was dead. But the truthful Yudhiṣṭhira would not lie. So an elephant named Aśvatthāmā was killed first, and Yudhiṣṭhira could now say, honestly, "Aśvatthāmā is dead" — adding, under his breath, "the elephant." The qualifier was drowned out. Droṇa, believing his son gone, set down his weapons and was killed.
Every word was true. The context was buried. A war turned on the difference.
That is the whole point, compressed into one of the most consequential sentences ever spoken. The truth was told. The context was withheld. And the outcome bent entirely around the missing half.
It's a striking thing to land on from a screenshot about a car wash — but it's the same shape. An answer can be completely true and completely misleading at once, depending only on the context riding alongside it. The machine that told you to walk wasn't lying. Neither was Yudhiṣṭhira. In both, everything hung on what went unsaid.
For a circle of leaders, that's the take-home. The quality of what comes back to you — from a model, from a team, from a single decisive moment — is set by the quality and the completeness of what you put in. Withhold the context, even truthfully, and you can win the wrong war.