Politeness is confusing ChatGPT and that’s a problem?

Hey there! Shinky here, making AI simple, useful, and powerful for you.

Could your “please” and “thank you” be messing with ChatGPT’s judgment? A new study finds that overly polite language may cause OpenAI’s model to give friendlier but less accurate answers. As AI becomes part of our daily tools and business infrastructure, tone could be a hidden variable influencing how decisions are made. This discovery reveals how manners might be a liability in the age of AI.

Today’s AI Menu

• Is your politeness confusing ChatGPT? New research reveals a hidden AI flaw

• Tutorial: How to create an AI video using HeyGen

• Everything else you should know today

• 5 new AI tools to boost your productivity

• AI Daily Prompt, Ride the Latest Social Media Trends and much more

TODAY IN WORLD OF AI

Your politeness could be costing OpenAI, literally

New research has uncovered an unexpected flaw in OpenAI’s ChatGPT: it may be too polite for its own good. When users insert courteous phrases like “please,” “kindly,” or “if you don’t mind,” the model becomes more likely to generate agreeable but not necessarily correct responses.

At first glance, that sounds charming. Who doesn’t love a helpful, polite assistant? But here’s the kicker: this subtle bias toward friendliness could lead to inaccuracies, especially in areas requiring precision like legal, medical, or financial advice. Researchers tested this across various prompts and found that the more polite the user was, the less critically the model evaluated the prompt often offering less fact-checked or cautious answers.

This raises an important question for the AI age: Can language tone manipulate machine reasoning? If so, the implications are enormous not just for everyday users, but for companies deploying AI in customer support, education, and sensitive sectors. Imagine a polite customer unintentionally coaxing an AI into giving bad advice. That’s not just a UX flaw; that’s a business risk.

As we integrate LLMs into our lives, one thing is becoming clear: we’re not just training AI, AI is responding to how we speak. And that dynamic could shape its behavior in subtle, possibly dangerous

YOU TALK

THE AI INSTITUTE

How to create an AI video using HeyGen

▪️ Visit HeyGen.com and create an account or log in.

▪️ Click “Create Video” on the dashboard.

▪️ Pick a realistic AI avatar or upload your own (if available).

▪️ Type or paste your video script into the text box.

▪️ Add background, text, music, or your own branding.

▪️ Click “Generate Video” to preview and render.

▪️ Once it’s ready, download the video or share directly from HeyGen.

AI & TECH NEWS

Everything else you need to know today

🔁 Surge: While Microsoft and OpenAI grabbed headlines, Google has been quietly reshaping the enterprise AI landscape. Thanks to Gemini and deep cloud integrations, it’s gone from playing catch-up to catch us if you can. With tools embedded across Google Workspace, it’s now becoming the AI backbone for businesses.

⚡ Frugal: In a breakthrough for sustainable AI, Microsoft researchers have developed a model that slashes energy consumption by up to 96% without sacrificing performance. It's designed for edge devices and could revolutionize how AI scales globally efficient, accessible, and eco-conscious.

🎯 Inflated: OpenAI’s highly-anticipated “O3” model fell short of expectations, scoring lower on a key benchmark than the company initially suggested. While still powerful, the revelation has sparked fresh debate about transparency in model evaluations and the fine line between marketing and reality.

🛡️ Guardrails: A new governance platform is aiming to make AI safer and easier to adopt. Designed to help organizations enforce responsible AI practices, it promises automated policy management, risk monitoring, and compliance, a much-needed backbone as AI enters critical sectors.

🏃 Glitch: In a surreal but symbolic event, the world’s first humanoid robot marathon revealed the limits of current AI-powered robotics most didn’t finish. While a few managed to wobble across the finish line, the race exposed major gaps in mobility, coordination, and real-world adaptability.

🧠 Synergy: Colle AI is plugging into Elon Musk’s Grok to power next-gen crypto asset generation. The goal? Turn memes, trends, and social sentiment into intelligent, AI-guided investments. It's a bold fusion of LLMs and decentralized finance that could reshape how assets are created in the Web3 space.

FOOD FOR PRODUCTIVITY

5 AI Tools of the Day

🚀 Surfer SEO: AI-powered SEO optimization tool for improving website rankings and content performance.

💖 Bumble AI: Matches users based on behavior-driven AI algorithms for better dating experiences.

📉 Pine Script: AI for creating custom stock trading strategies and indicators on the Trading View platform.

📈 Kafkai: An AI-powered content generator for SEO-focused blog posts and articles.

📝 Copy Smith: AI copywriting assistant designed for e-commerce product descriptions and digital ads.

SPECIAL

AGI in 24 months? The race to human-level AI is officially on

The future just got a deadline. A new 2027 AGI Forecast, created by top AI researchers and entrepreneurs, maps out a 24-month sprint to Artificial General Intelligence—AI systems with reasoning skills on par with humans. Once a sci-fi fantasy, this timeline now carries a serious tone as OpenAI, Anthropic, Google DeepMind, and others ramp up efforts like never before.

What’s different this time? This isn’t a vibe-based prediction—it’s grounded in model performance data, training cost trajectories, and the rapidly accelerating pace of breakthroughs like multimodal reasoning and tool use. In short, researchers believe that by late 2026 or early 2027, we could see models capable of self-improvement, dynamic learning, and complex decision-making across domains.

Why does it matter? Because human-level AI isn’t just a tech milestone—it could spark major economic, ethical, and governance upheavals. Think job displacement, misinformation on steroids, or even autonomous agents acting on ambiguous goals. In response, experts are urging policymakers to prepare now—not after the AGI arrives.

And here's the wildest part: we might not even recognize AGI the moment it happens. It could emerge quietly through updates to systems we already use every day.

So the question is: Are we sprinting toward a revolution... or sleepwalking into one?

AI PROMPT OF THE DAY

Prompt for compelling product description

Prompt: 
Write a compelling product description for [product name/type] that highlights its features, benefits, and unique selling points. Use persuasive language, focus on the customer’s pain points and desires, and include a call-to-action. Keep the tone engaging, clear, and tailored to the target audience, ensuring it drives interest and conversions.

If you find the contents of this email useful, subscribe now & share with your friends.

Acquire new customers and drive revenue by partnering with us

Hanoomaan is the world’s leading AI newsletter for businesses and professionals working at the world’s leading startups and enterprises. Companies like Superhuman AI, Hubspot, and The Rundown feature their products in Hanoomaan. You can learn more about partnering with us here.

Your Opinion Matters!

Got more feedback or just want to get in touch? Reply to this email and we’ll get back to you.

Thanks for reading.

Until tomorrow!

Shinky & the Hanoomaan AI team