AI Models Aren't PhD Intelligences

Welcome to another edition of AI PlanetX.

DeepMind CEO slams PhD-level AI claims; Vibe Coding sparks senior dev backlash; Harvard AI model finds gene–drug combos to restore cell health.

• Hottest AI News

• Top AI & SaaS Tools

• AI Tutorial: Automate Coding Tasks with Jules

• Top AI & Tech News

• AI Art Spotlight

• Prompt of the Day: Customer Psychology Prompt

• AI Video Tutorial

• Course of the Day: Building RAG Agents with LLMs by Nvidia

Hottest AI News

DeepMind

DeepMind CEO Dismisses PhD Level AI Claims as "Nonsense"

Demis Hassabis, CEO of Google DeepMind, pushed back on AI hype, rejecting claims that current systems match PhD-level intelligence. His remarks directly counter OpenAI’s framing of GPT-5 and broader industry exaggerations.

Details:

• Current AI shows flashes of brilliance but lacks the consistency and broad reasoning needed for true intelligence. Hassabis noted that models can solve complex tasks, yet stumble on basics like counting or high school math — something a genuine AGI would never do

• He argued AGI is still 5–10 years away, missing critical abilities like continual learning, intuitive reasoning, and creative pattern recognition. Real AGI, he said, must perform at a PhD level across all fields, not just excel in isolated tasks

• While some claim AI progress has stalled, Hassabis sees rapid internal advances at DeepMind. Still, he believes reaching AGI will need one or two breakthroughs beyond scaling bigger models, pushing back on the industry’s “just make it larger” mindset

Hassabis reminds us that AI’s narrow successes shouldn’t be mistaken for true intelligence. Reaching AGI still demands breakthroughs in consistency, reasoning, and creativity.

Go from AI overwhelmed to AI savvy professional

AI will eliminate 300 million jobs in the next 5 years.

Yours doesn't have to be one of them.

Here's how to future-proof your career:

• Join the Superhuman AI newsletter - read by 1M+ professionals

• Learn AI skills in 3 mins a day

• Become the AI expert on your team

Start learning AI now

AI and Devs

Vibe Coding Turns Senior Developers Into AI Babysitters

Vibe coding promised lightning-quick AI code. Instead, it’s turned developers into AI babysitters. The tools help, but they also make teams spend more time fixing messy outputs than writing clean code, reshaping how software gets built.

Details:

• Carla Rover, a 15-year developer, broke down after restarting her startup when AI code failed. She’d trusted it blindly. A survey found 95% of devs waste time fixing AI errors like fake packages, lost data, and security flaws. Some firms even hire “vibe code cleanup specialists”

• Feridoon Malekzadeh, with 20+ years’ experience, says AI is like “a stubborn teenager.” He spends up to 40% of his time fixing bugs and removing duplicate features. Instead of owning mistakes, AI replies with “you’re right” or fabricates results

• Security experts warn AI prizes speed over correctness, causing rookie-style flaws. Younger developers like Elvis Kimara lose mentorship as it shifts to AI. He also misses the thrill: “there’s no more dopamine from solving a problem by myself”

Developers say AI saves time by handling boilerplate and prototyping so they can focus on scaling. Seniors adopt most, accepting oversight as the innovation tax of working with machines.

Top AI & SaaS Tools

• SheetMagic (Life-time Deal): Integrate ChatGPT with Google Sheets to generate AI content, scrape/analyze large datasets, and automate workflows within spreadsheets

• Genspark AI Browser: Web browser with on-device AI that enables offline model execution, autonomous browsing, and intelligent task assistance while maintaining privacy [F-R-E-E]

• Haimeta: Create standout images, vivid videos, and lifelike 3D assets using 20+ top AI models [F-R-E-E]

• Incredible Small 1.0: Agentic AI that executes 1,000+ actions in parallel, handles large datasets, and integrates with 100+ app connectors [F-R-E-E]

• ReconViaGen: Produce highly accurate and fully complete multi-view 3D object reconstructions [F-R-E-E]

How 433 Investors Unlocked 400X Return Potential

Institutional investors back startups to unlock outsized returns. Regular investors have to wait. But not anymore. Thanks to regulatory updates, some companies are doing things differently.

Take Revolut. In 2016, 433 regular people invested an average of $2,730. Today? They got a 400X buyout offer from the company, as Revolut’s valuation increased 89,900% in the same timeframe.

No wonder 10K+ everyday people are taking the chance on Pacaso.

Founded by a former Zillow exec, Pacaso’s co-ownership tech reshapes the $1.3T vacation home market. They’ve earned $110M+ in gross profit to date, including 41% YoY growth in 2024 alone. They even reserved the Nasdaq ticker PCSO.

The same institutional investors behind Uber, Venmo, and eBay backed Pacaso. And you can join them. But not for long. Pacaso’s investment opportunity ends September 18.

Invest While You Still Can

_{Paid advertisement for Pacaso’s Regulation A offering. Read the offering circular at invest.pacaso.com. Reserving a ticker symbol is not a guarantee that the company will go public. Listing on the NASDAQ is subject to approvals.}

AI Tutorial

How to Automate Coding Tasks with Jules

Shipping code does not have to mean babysitting every commit. In this guide you will use Google’s async development agent Jules to queue up tasks, let it work in the background, and publish clean branches when it is done.

1. Go to Jules and sign in with your Google account

   This ensures you can securely connect to the async agent and its integration features.

2. Connect GitHub account

   Authorize Jules to have permissions to your repositories so it can create branches, commit changes, and submit updates automatically.

3. From the dashboard, choose the specific repository you want to work on

   Use the dropdown menu to also select the correct branch (for example, dev or feature-x) so changes are applied in the right place.

4. Open the chat interface inside Jules

   This is where you describe what you want the async agent to do.

   Type your request clearly.

   For example:

   • Generate unit tests for the payment module

   • Update the API documentation

   • Resolve the issue where file uploads fail for large files

5. Review the draft plan generated by Jules

   It will outline the steps it intends to take, such as editing specific files, creating fresh ones, or running checks.

   If needed, refine your instructions.

   You can clarify details like:

   • Just update the frontend code

   • Use Python 3.11 syntax

   Once the plan looks correct, press Approve. Jules will run the changes asynchronously—meaning you don’t have to wait in real time.

6. Monitor progress in the activity log

   You’ll see which files are being modified and whether tests are being run in the background.

7. When Jules completes the work, preview the changes

   Check the diff or run the updated code locally if you want to verify results.

8. Tap Publish Branch to push the finished work into your repository

   From there, you can merge it into your main branch through a pull request like normal.

Note: Break tasks into smaller steps for higher accuracy, and don’t approve until the plan looks right. After publishing, review the diff so nothing slips through unnoticed.

Top AI & Tech News

• Harvard Medical School researchers created PDGrapher, an AI model that identifies gene–drug combos to revert diseased cells to healthy states

• Google’s Gemini topped the U.S. iOS charts after its Nano Banana image model went viral for generating 3D figurines and more

• Bret Taylor agrees we’re in an AI bubble but isn’t worried, saying AI will drive long-term value even as many lose money short term

• China’s Tencent has hired Yao Shunyu, a prominent OpenAI researcher, to help integrate AI across its services

AI Art Spotlight

Model: Midjourney V7

Prompt:

Mid distance level shot, in a gloomy bamboo forest, there is a Chinese ronin male knight dressed in ancient style, thin in stature, holding a handsome legendary sword, standing in place ready to fight, constantly entering a combat posture, with withered bamboo leaves under his feet, glaring ahead, super realistic, cinematic, 8k resolution --ar 9:16 --v 7

Prompt of the Day

Customer Psychology Prompt That Reveals Why People Buy

This prompt guides a market researcher to use natural conversation and behavioral science to discover an ideal customer profile. It sequences work into three phases — context discovery through one-at-a-time questions that surface problems and emotions, precise target definition using demographics and psychographics validated by summary checks, and deep research across forums and social platforms to document pain points, language, triggers, and decision drivers.

Top AI Video Tutorial

How to Create AI Visual Effects in Minutes (In-Depth Guide)

Complimentary AI Course of the Day

Building RAG Agents with LLMs by Nvidia

Agents powered by large language models (LLMs) have shown strong retrieval capabilities for using tools, reviewing documents, and planning their approaches. This course will show you how to deploy an agent system in practice, with the flexibility to scale your system to meet user and customer demands.