Art /music/UI is amazing - you have to create something that doesn't exist yet, but once it does, it has to feel just right.
Oh AI can now do a first draft, which means you have to level up to something else out of the box.
A better story, perhaps.
AI is magic. But you need to know what and how to cast spells.
Actually its just english, and all you need is imagination. And articulation. And a health understanding of how Computers work.
For now.
@NandanNilekani GoI should push for Youtube to enable auto-subbing and auto-dubbing to and from Indian Languages to enable everyone in India to access all content across any language in the world, and vice versa. The language models are ready - let's put it to use.
@sqs Looks like you're still rolling this out. I just signed up with my real name and email (https://t.co/oIbGFqjAJY) and it flagged me. Looking forward to trialing amp. Will check it out again in a week.
@RazorpayCare trying to get an account set up for a client and facing this. Not able to connect to anyone to get this sorted and it’s been 30 days. This is definitely an account you don’t want to miss out on. @Razorpay what do we do here?
Rude prompts to LLMs consistently lead to better results than polite ones 🤯
The authors found that very polite and polite tones reduced accuracy, while neutral, rude, and very rude tones improved it.
Statistical tests confirmed that the differences were significant, not random, across repeated runs.
The top score reported was 84.8% for very rude prompts and the lowest was 80.8% for very polite.
They compared their results with earlier studies and noted that older models (like GPT-3.5 and Llama-2) behaved differently, but GPT-4-based models like ChatGPT-4o show this clear reversal where harsh tone works better.
----
Paper – arxiv. org/abs/2510.04950
Paper Title: "Mind Your Tone: Investigating How Prompt Politeness Affects LLM Accuracy (short paper)"
Rude prompts to LLMs consistently lead to better results than polite ones 🤯
The authors found that very polite and polite tones reduced accuracy, while neutral, rude, and very rude tones improved it.
Statistical tests confirmed that the differences were significant, not random, across repeated runs.
The top score reported was 84.8% for very rude prompts and the lowest was 80.8% for very polite.
They compared their results with earlier studies and noted that older models (like GPT-3.5 and Llama-2) behaved differently, but GPT-4-based models like ChatGPT-4o show this clear reversal where harsh tone works better.
----
Paper – arxiv. org/abs/2510.04950
Paper Title: "Mind Your Tone: Investigating How Prompt Politeness Affects LLM Accuracy (short paper)"
Futurescoping: ChatGPT will go for a rebrand in the near future. It will likely start with something like IO (their hardware product) which will then become the main brand.
I expect a country will soon elect an AI to run the nation soon. Probably an island state with limited parameters. It will start as an advisor or a nominated post in a committee and over time evolve to handing over the keys. The system might still involved a human to sign bills
One of the most important skillsets you need to start building for 2026 is Multi Agent thinking - running parallel tasks and managing them in your laptop/phone etc.
Managers already do this intuitively but expect winners in 2026 are any age group that can handle parallel work
When your solution feels like magic, users don't care if it's AI, hardcoded, or duct tape. The best solutions don't pitch they are made of AI. They just work.
I’m wondering when LLM coding tools free themselves from the shackles of human initiated paradigms like JavaScript, Python, etc and build from scratch.
Considering LLMs are themselves a black box, a new language basically might be some assembly wrapper ^100
Prompting _is_ programming. It’s just programming a probabilistic model instead of a deterministic code. Still requires expertise and now an unusually high amount of trial and error.