Giving ChatGPT the Ability to Execute Tasks

Leo Huang
3 min readNov 11, 2023

“The relationship between human intelligence and artificial intelligence (HI + AI) will necessarily be one of symbiosis. The challenge and potential of exploring this co-evolutionary future is the biggest story of the next century and one in which a closeness in development velocity is a necessity.”

- Bryan Johnson

I n the realm of AI, a new dawn is coming, one of AI Agents. AI Agents essentially are your LLMs (ChatGPT, Bard, You.com), but we give it the ability to execute commands. From, web searches to creating documents, the author makes the only limitation AI.

Credit: Unsplash

The integration of AI Agents into various facets of our lives holds the promise of streamlining and enhancing our daily experiences. Imagine having an AI Agent capable of understanding your commands and executing tasks seamlessly. From scheduling appointments, drafting documents, and analyzing data, to providing personalized recommendations, these AI Agents are set to become indispensable aides, effectively augmenting human productivity and efficiency across industries.

But this can be a reality, and it can be done so fairly simply if you’re willing to work with it:

Getting Everything We Need

We have to have a few things installed before we can actually do this so we can properly run our program:

  • Docker — Docker is a platform that uses containers to create and deploy applications. These containers are lightweight, portable, and operate in isolation, sharing the host system’s OS kernel for efficiency. Docker simplifies application development by using images as templates to package applications and their dependencies consistently, enabling portability and ease of deployment across various environments.
  • Python — Python is a high-level programming language known for its simplicity and readability. It emphasizes clean, concise, and easily understandable code, making it popular among developers for various applications. Python supports multiple programming paradigms and has a vast array of libraries and frameworks, facilitating tasks ranging from web development, data analysis, artificial intelligence, machine learning, to scripting and automation. Its versatility, ease of use, and community support make Python an excellent choice for both beginners and experienced programmers.
  • Git — Git is a version control system used in software development to track file changes, allowing collaboration among multiple developers on projects. It maintains a history of modifications, supports branching for parallel work, and enables easy synchronization and merging of changes, making it essential for managing code and team collaboration.

All this, along with some terminal, we can finally start working on this.

Cloning and Setup

Now, we can either make this from scratch, or we can participate in AutoGPT’s ongoing work, maybe even contribute who knows.

So, let’s do:

git clone -b stable https://github.com/Significant-Gravitas/AutoGPT.git

And we have most of our work done for us, but we have to change a few things.

Firstly, we need an API key, to use ChatGPT, so make sure that you have one, if you need help, check out here.

Next, we’re going to go to the .env.template file and locate the OPENAI_API_KEY, remove the hashtag in front of it, then add our OpenAI API key in there.

Now run: ./run

and boom, we’re done.

So now we’ve gotten to the point where we can run the agent and create our own AI Agent. The possibilities are endless, and there is no reason as to why we can’t add more commands, or add more APIs.

Conclusion

So that’s how we give ChatGPT the ability to work all on its own to execute our tasks. But the real fun begins when we take a look at the code and we change it for the better. There are so many upgrades we can make to this such as making more commands, prompting the AI better, or even changing the code to be more efficient. Whatever it is though, I hope that we now can test our hands on some of the most exciting technology yet!

Further Readings & Source

  1. https://www.youtube.com/watch?v=-DlXcqpheIg
  2. https://github.com/Significant-Gravitas/AutoGPT/tree/master
  3. https://openai.com/

Hey👋, if you’ve enjoyed this article, or just want to discuss AI with me. Feel free to reach out to me and I’d love to have a chat.

--

--