Why has Musk's 'Macrohard' AI project stalled?
What's the story
Elon Musk's ambitious AI agent project, Macrohard, has hit a roadblock due to leadership changes and the suspension of a data project involving 600 contractors. At the same time, Musk's other company Tesla is ramping up its own AI agent project called "Digital Optimus," according to Business Insider.
Project overview
Musk's Macrohard AI project
Musk announced the tongue-in-cheek project, Macrohard, in August. He had said that since software companies like Microsoft don't manufacture physical hardware, they could be simulated entirely with AI. The goal was to create an AI white-collar worker. However, the project has struggled with leadership changes and scaling up since its inception.
Personnel changes
Leadership changes and development pressure
Two leaders of the Macrohard project left the company in February. Shortly after, Musk announced xAI cofounder Toby Pohlen would lead the project. However, Pohlen announced his exit just 16 days later amid pressure from Musk regarding development progress. In recent weeks, some employees were told their work would shift to Tesla's Autopilot team along with some of Macrohard's computing capacity.
New initiative
Tesla's digital Optimus AI agent
Meanwhile, Tesla has been working on an internal AI agent called "Digital Optimus." The digital version is intended to serve as an AI agent capable of performing tasks on a computer. Unlike traditional models that rely on screenshot analysis, the team is focusing on real-time control methods for this project. This approach is similar to how Tesla's Full Self-Driving system processes live video data to help its AI learn and navigate in real time.
Project status
Data annotation project paused last month
The data annotation project for Macrohard that involved over 600 AI tutors was paused last month. The workers were asked to screen record their work and leisure activities to train the AI how to emulate their actions as a digital agent. A project lead had informed them that researchers had "discovered many flaws within the model, and would like to make some changes to our model and the way we collect data."