AI Startup Aims to Automate Every Job

Welcome to another edition of AI PlanetX.

Visionary Founder Unveils Ambitious AI Startup to Replace All Human Jobs; OpenAI's Advanced Reasoning Models Struggle with Accuracy.

• Hottest AI News

• Top AI & SaaS Tools

• Top AI & Tech News

• Interesting Uses of AI: OpenAI's Guide to Build Agents

• AI Video Tutorial

• F-R-E-E AI Course of the Day: ChatGPT 101 - Complete Beginner's Guide

Hottest AI News

Mechanize

AI Pioneer Launches Startup to Replace All Human Workers Worldwide

A controversial Silicon Valley startup, Mechanize, launched Thursday with the bold goal of automating all human work globally. Founded by AI researcher Tamay Besiroglu, its audacious mission has sparked debate over whether it's serious or satire.

Details:

• Mechanize’s mission is to fully automate all work and the global economy, with a market potential of $18 trillion annually in the US and over $60 trillion globally

• The launch has sparked backlash, even from Besiroglu’s own research group, Epoch. Critics argue it threatens jobs and tarnishes Epoch’s reputation as an unbiased AI research institute

• Despite the criticism, Mechanize has secured funding from top investors like Nat Friedman, Daniel Gross, and Jeff Dean. Besiroglu argues full automation will boost economic growth and living standards

This isn't the first controversy surrounding Besiroglu's organizations. Epoch previously faced criticism for not transparently disclosing its relationship with OpenAI when releasing AI benchmarks.

Get Over $6K of Notion Free with Unlimited AI

Running a startup is complex. That's why thousands of startups trust Notion as their connected workspace for managing projects, tracking fundraising, and team collaboration

Apply now to get up to 6 months of Notion with unlimited AI free ($6,000+ value) to build and scale your company with one tool. 

Get Free Notion + AI

OpenAI

OpenAI's New Reasoning Models Hallucinate More Than Predecessors

Surprisingly, OpenAI's o3 and o4-mini models hallucinate more often than older versions, highlighting ongoing challenges in AI development despite their impressive capabilities in certain areas.

Details:

• OpenAI's testing shows o3 hallucinated 33% of the time on PersonQA, double the rate of earlier models. o4-mini fared worse, with 48% hallucinations, much higher than non-reasoning models like GPT-4o

• The company admits it doesn’t understand why hallucinations rise with increased reasoning, acknowledging a knowledge gap and the need for more research

• Researchers found issues like models inventing processes. For example, o3 falsely claimed to run code on a MacBook Pro outside ChatGPT, and users reported problems like non-functional website links

OpenAI is focused on improving reliability as the shift to reasoning models makes addressing hallucinations critical for business use. Web search integration shows promise, with GPT-4o hitting 90% accuracy on SimpleQA.

Top AI & SaaS Tools

• Castmagic (Life-time Deal): AI that turns audio and video into ready-to-use content—transcripts, summaries, and repurposed formats—for podcasts, meetings, sales, coaching, and more [145+ five-star reviews]

• InstantCharacter: Personalize any character by uploading an image, describing your vision in a text prompt, and generating a unique, shareable creation [F-R-E-E]

• Kling: Edit any video by swapping, adding, or removing content, even changing one actor for another in any scene (see this tutorial) [F-R-E-E]

• Dream 7B: Unlike traditional AI that generates text word-by-word, it produces and refines text all at once, matching top models in performance [F-R-E-E]

• EverTutor: AI voice tutor personalizes interactive lessons to your learning style, providing feedback and tailored GRE prep in real time [F-R-E-E]

Top AI & Tech News

• Some ChatGPT users are unsettled by the chatbot's habit of addressing them by name unprompted, calling it "creepy" and "unnecessary"

• Fudan University researchers have created "PoX," the fastest non-volatile flash memory, writing a bit in 400 picoseconds—about 10,000 times faster than current flash memory

• Google DeepMind proposes an AI development approach called "streams," which enables AI agents to learn from continuous interaction with the environment rather than static, human-labeled data

• Twenty-one humanoid robots participated in the Yizhuang half-marathon in Beijing, marking the first time machines raced alongside humans in a 21-km course

Interesting Uses of AI

AI Art Spotlight

Model: Midjourney V7

Prompt:

A tracking shot captures a young, long-haired Japanese woman riding a cheetah through the forest at breakneck speed. She is wearing a white blouse. She is laughing with excitement and clinging to the cheetah. The cheetah is running at breakneck speed, raising a cloud of dust. The camera shakes at the intense speed. --ar 9:16 --raw --profile mxfxok9 --v 7

AI Guide of the Day

OpenAI has just published a comprehensive, 34-page practical guide on building agents. This guide offers step-by-step instructions and best practices for anyone interested in developing intelligent agents!

Top AI Video Tutorial

7 Mind-Blowing Use Cases of NotebookLM

Complimentary AI Course of the Day

ChatGPT 101: Complete Beginner's Guide and Masterclass

This course covers how to train ChatGPT to write emails, design a structured online course outline, and identify the best Google Chrome extensions that work well with ChatGPT. You’ll also learn how to use AI to boost your productivity and streamline your workflows for better efficiency and results.