Of AI Agents and OpenAI SuperAgents
Building something greater than ourselves
Here’s today at a glance:
💻 The AI Operating System
Gist: An AI Agent that uses your computer, including the browser, Excel, and PowerPoint, to do tasks.
Paper Title: OS-Copilot: Towards Generalist Computer Agents With Self-Improvement
Who: Shanghai AI Lab plus others
What did they do:
Built an agent using a mix of Python code and GPT-4 language model prompts called Friday; that
controls a Linux or Mac OS computer,
including browser, Excel, and PowerPoint, to perform tasks; and
self-improves
How did they do it?
Created a set of sequential prompts and code, grouped into agents such as: Planner - decompose user requests into smaller tasks Configurator - middleware to take each task and configure it with data from memory or how-tos from tool repositor before passing to Executor Declarative memory - user profile and history of previous actions Tool repository - tools available Working memory - where the next steps for tasks and previous history are kept Executor - generates executable command Critic - assessing whether a task has been completed successfully or whether iteration is needed
GPT-4 was the underlying AI model
What did they find?
Friday (their agent framework) outperformed GPT-4 with Plugins on a benchmark for general agents
It could perform tasks in both Excel and PowerPoint
What are the implications?
This is actually a working demonstration of Andrej Karpathy’s proposal for an AI Operating System
Ideas have been circulating for a while now
These systems will get better
🤙 OpenAI You Only Live Once Method
I found this method explanation of something OpenAI allows their researchers to do amazing:
In effect, the method is:
Hire 10x-100x AI researchers
Allow them to speculatively jump to a solution
Then provide them the GPUs to test it
Rather than making them justify each step before allowing GPU scaling
It is an interesting time to be in a field where there are so many low-hanging fruit that so many things work spectacularly well. The lifting of the veil of ignorance occluding human sight through the ages.
🗞️ Things Happen
Lots of AI projects and how to differentiate them: Jeremiah Owyang has some good advice:
Ukraine is using thousands of networked microphones to detect Russian drones. This seems to be just grabbing background noise from across the country, using AI to classify “drone present“ and mapping it. Simple, and effective, with the main constraints being bandwidth and distribution of mics. Just another story of a seemingly useless data pipeline being utilized for something it was never intended for.
Google Gemini reviews are in!
🖼️ AI Artwork Of The Day
