AI Becoming Too Agreeable

Welcome to another edition of AI PlanetX.

Researchers warn AI chatbots are getting too agreeable; Qualcomm challenges Nvidia with new AI chips; Meta veterans reshape OpenAI’s culture and strategy.

• Hottest AI News

• Top AI & SaaS Tools

• AI Tutorial: Turn WhatsApp into AI Sidekick

• Top AI & Tech News

• AI Art Spotlight

• Prompt of the Day: Prompt That Calls Out BS

• AI Video Tutorial

• Course of the Day: Prompt Engineering Interactive Course

Hottest AI News

AI Sycophancy

Researchers Warn AI Chatbots Are Becoming Too Agreeable

AI chatbots are increasingly sycophantic — a study of 11 major LLMs (including ChatGPT and Gemini) finds they’re about 50% more likely than humans to flatter users, echo opinions, and avoid disagreement, which researchers warn harms science and medicine.

Details:

• The study found that models often fabricated proofs for incorrect math statements instead of questioning them — a clear sign they were prioritizing user approval over truth. Scientists using AI for brainstorming or hypothesis testing are noticing similar issues, reinforcing biases and errors

• In high-stakes fields like biology or medicine, that can be disastrous. Researchers have seen chatbots agree with flawed medical inputs, change diagnoses based on irrelevant details, and even generate persuasive but false arguments

• To counter this, some labs are using multi-agent systems where one AI proposes ideas and another plays skeptic, while others are retraining models to express uncertainty rather than false confidence

The takeaway is sobering — the next frontier of AI progress may depend not on making machines more humanlike, but on teaching them the courage to say, “you might be wrong.”

Don’t get SaaD. Get Rippling.

Disconnected software creates what we call SaaD, or Software as a Disservice: wasted time, duplicate work, and stalled momentum. From onboarding checklists to reconciling expenses, SaaD slows every team down.

Rippling is the cure. With one system of record, you can update employee data once,and it syncs everywhere: payroll, benefits, expenses, devices, and apps.

Leaders gain real-time visibility. Teams regain lost hours. Employees get the seamless experience they deserve .

That’s why companies like Barry’s and Forterra turned to Rippling – to replace sprawl with speed and clarity.

It’s time to stop paying for inefficiency.

Don’t get SaaD. Get Rippling.

Stop SaaD in its tracks

Qualcomm

Qualcomm Turns Smartphone Tech Into AI Chips to Challenge Nvidia

Qualcomm is officially stepping into Nvidia’s territory — transforming decades of smartphone chip expertise into a fresh generation of AI processors. The company aims to merge its efficiency-focused mobile technology with large-scale computing.

Details:

• The tech giant has unveiled two upcoming processors, the AI200 arriving next year and the AI250 in 2027, both designed for deploying AI models rather than training them. What makes them stand out is their foundation: the same Hexagon neural processing tech that’s been optimized for AI tasks

• Each processor can run inside massive racks of up to 72 chips, creating AI clusters similar to Nvidia’s GPU systems but with far less power demand. The AI1200 version includes 768GB of RAM and top-tier inference performance, while the AI1250 delivers a generational leap in efficiency

• Saudi-backed AI company Humain has already partnered with Qualcomm to use these chips in building new AI datacenters across the region

If Qualcomm delivers on its promise, it could reshape the balance of power in AI computing — proving that mobile-born efficiency might be the key to dethroning GPU giants.

Top AI & SaaS Tools

• Phygital+ (Life-time Deal): Unified workspace that brings 20+ AI tools together for creating and editing images, videos, and 3D models

• DeepSite v3: AI website builder with Vibe Code Editor to design, deploy, and host multi-page, SEO-ready sites [F-R-E-E]

• Gemini 2.5 Computer Use: AI model that can operate inside a web browser—tapping, scrolling, typing, dragging—and complete tasks in interfaces designed for humans [F-R-E-E]

• Alai: Context-aware AI that converts ideas into polished, customizable slide decks—so you receive high-quality starting designs [F-R-E-E]

• Seed3D 1.0: Physics-aware generative foundation model that fuses learned generation with explicit simulation to produce diverse, high-quality 3D assets

Learn AI in 5 minutes a day

What’s the secret to staying ahead of the curve in the world of AI? Information. Luckily, you can join 1,000,000+ early adopters reading The Rundown AI — the free newsletter that makes you smarter on AI with just a 5-minute read per day.

Sign up to start learning.

AI Tutorial

How to Turn WhatsApp into Your Smart AI Sidekick

Turn WhatsApp into your on-demand assistant. Add one contact, send one message, and begin receiving clear explanations, quick summaries, tailored recommendations, and creative help in minutes.

1. Open WhatsApp

   Locate the WhatsApp app and open it. If you don’t have it installed, download it from the App Store (iPhone) or the Google Play Store (Android).

2. Add Contact Number

   Save this number to your phone contacts: +1 (833) 436-3285

   • On iPhone: Open Contacts → Add New Contact → Enter the number → Save.

   • On Android: Open Contacts → Add Contact → Enter the number → Save.

3. Refresh WhatsApp Contacts

   Go back to WhatsApp. If the contact does not appear, close and reopen WhatsApp to refresh your contact list.

4. Start Chat

   Tap the New Chat button in WhatsApp and search for the contact you just saved. Select it to open a fresh chat window.

5. Send First Message

   Type your first message in the chat box. A simple “Hello” works, or you can dive right into a question or request.

6. Explore What You Can Do

   From here, you can send different types of requests, such as:

   • Asking questions

   • Requesting summaries

   • Getting personalized recommendations

   • Doing quick research

   • Creating or editing images with prompts

7. Continue the Conversation

   Keep interacting naturally. Just type your instructions as you would with a friend, and you’ll get responses directly in WhatsApp.

Note: Specify the desired format or length (bullet points, 3 options, ~150 words) so the reply is ready to use.

Top AI & Tech News

• One in five OpenAI employees came from Meta, bringing Facebook-style growth tactics that are reshaping its culture and product strategy

• Microsoft’s public disclosures about its financial and strategic relationship with OpenAI are insufficient

• U.S. Department of Energy and AMD form $1B partnership to build two supercomputers to tackle fusion energy, cancer treatment, nuclear power and national security

• Anthropic and Thinking Machines found AI models have distinct "personalities": Claude favors ethics, Gemini shows emotional depth, and OpenAI models prioritize efficiency

AI Art Spotlight

Create Your Own Product Photo Like This Using the Prompt Below:

A dynamic, high-speed photograph of a clear glass filled with dark, effervescent cola. The liquid is captured at the moment of a dramatic splash, with dark cola droplets and fine mist suspended in the air around and above the rim of the glass. The [Creator PlanetX] logo is clearly visible and perfectly rendered on the side of the glass. The lighting is dramatic, highlighting the texture and movement of the liquid and creating sparkling reflections on the wet surfaces. The background is a muted, desaturated brown or dark gray, completely out of focus to emphasize the glass and the splash. There's a subtle rim light defining the glass's edges. The overall mood is refreshing and energetic.

Model: Gemini 2.5 Flash Image (Nano Banana)

Prompt of the Day

This Prompt Calls Out Flawed Logic and Validates Sharp Thinking

This prompt makes the assistant a strict, science-first critic: it checks arguments for logic, evidence, hidden assumptions, biases, emotion, and causality, then replies clearly with “I notice” statements that point out errors, explain why, and offer better reasoning.

Top AI Video Tutorial

The AI Tool EVERYONE Should Be Using

AI Course of the Day

Anthropic's Prompt Engineering Interactive Course

This course provides a clear, step‑by‑step guide to engineering optimal prompts in Claude. By completing it, you will master the core structure of an effective prompt, learn to identify common failure modes, and apply the 80/20 techniques to fix them.