Home > Technology peripherals > AI > body text

AutoGPT has exceeded 100,000 stars. This is the first article to systematically introduce autonomous agents.

WBOY
Release: 2023-04-28 16:10:16
forward
1493 people have browsed it

On GitHub, AutoGPT’s star count has exceeded 100,000. This is a new way of human-computer interaction: you don’t need to tell the AI ​​what to do first and what to do next, but just set a goal for it, even if it’s as simple as “creating the best ice cream in the world.” Similar projects include BabyAGI and so on. What does this wave of autonomous agents mean? How do they work? What will they look like in the future? How to try this new technology at this stage? In this article, Matt Schlicht, CEO and co-founder of Octane AI, goes into more detail.

Artificial intelligence can be used to complete very specific tasks, such as recommending content, writing copy, answering questions, and even generating photos that look like real life. You tell the AI ​​which task to complete, and it completes that task. It's very simple.

But what if you don’t want to help the AI ​​list all those tasks? What if you want a teammate more than just a tool? What if you want an AI to think for itself?

Imagine you created an AI tool that could set goals for it, even something as vague as “create the best ice cream in the world.” Make a to-do list, work on the to-do items, and add new ones as you progress. It then continues to repeat this process until it completes its goal.

This is exactly what “autonomous agents” do. They are among the fastest-growing trends among AI developers, but so far, most people are unaware of them. (As of this writing, no mainstream publication has written about autonomous agents. And only a handful have reported on it since its introduction, so if you're reading this... .... Then you are one of the first people to learn about it.)

What is an autonomous agent? Why are there huge opportunities behind them? How do they work? What will they look like in the future? How can I create or use it?

This is exactly the question I will answer for you.

「[INTELLIGENCE] Autonomous agents are often the natural end point of automation. In principle, agents can be used to automate any other process. It is not difficult to imagine that once these agents become highly sophisticated, Reliable, the degree of automation in various fields and industries will increase exponentially." ——Bojan Tunguz, NVIDIA machine learning expert

What is an autonomous agent?

Autonomous agents are programs empowered by AI. When given a goal, they are able to create tasks on their own, complete the task, create new tasks, reprioritize the task list, complete new top tasks, and repeat this process until the goal is achieved.

Please read the above description one more time, it is simple but crazy.

"Judging from the development trend of autonomous intelligence, everyone is expected to become a manager." - Yohei Nakajima, founder of BabyAGI

Autonomous agents can be designed to do anything from managing social media accounts and investing in the markets to creating the best children's books.

"Is this true? Can it be achieved now"

Yes, I know this sounds like science fiction, but These are real. If you know how to code, you can design one in minutes. And this is just the beginning.

"People often waste too much time doing tedious manual tasks. When computers can do these tasks, people can be freed to pursue more creative pursuits, or Do things that currently only humans can do. Autonomous agents will enable people to get more done in less time, and, over time, the amount of time people spend staring at screens will also be reduced!” — —Erica Brescia, Managing Director, Redpoint

The programming technology and AI required to realize autonomous agents are very realistic and extremely novel. Many open source projects, such as AutoGPT, BabyAGI, and Microsoft’s Jarvis, are popular in the AI ​​community and on Github.

In the first two weeks of creating the open source autonomous agent code base, nearly 100,000 developers have been building autonomous agents, optimizing them, and finding the upper limits of their capabilities, and this work has only just begun. Made in the weeks before these concepts were invented. Today, the number of developers using this technology is soaring at an ever-increasing rate.

「AI Agents will be everywhere.Billion-dollar companies will have a small deployment of AI agents. Developed by the team." ——Ben Tossell, founder of Ben's Bites AI Newsletter pytorch etc.

AutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

Auto-GPT Github's popularity is growing exponentially faster than any code base in history.

This is not science fiction. Many believe these autonomous agents are the beginning of true artificial general intelligence, or "AGI" - a term used to describe artificial intelligence that has gained sentience and become "alive."

"Autonomous agents may eventually commoditize all applications of factual knowledge. If access to factual knowledge also becomes universally available, creativity, emotion, and strategic vision etc., human qualities will become more precious and unique. However, knowledge is also likely to become increasingly proprietary, as individuals and companies seek to gain economic advantage in a world where the application of factual knowledge is commodified and human collective knowledge begins to stagnate. .” —Tony Hu, former acting director of emerging technologies for the FBI and co-founder of Bondoo AI It was installed in the browser and helped people order pizza.

You just say "Order a large, topping-free pizza delivered to One Vanderbilt from Dominos" and it orders itself.

HyperWrite's autonomous agent controls the browser to order pizza.

AutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

Or, check out this experiment from Stanford University and Google, where they created a virtual town of 25 autonomous agents and told one of them to come Organize a Valentine's Day party. This example is perhaps even more impressive.

The people simulated by these autonomous agents went about their days, talked to each other, formed new memories, and eventually most of them heard about the Valentine's Day party and ended up attending it .

## Picture source: "Generative Agents: Interactive Simulacra of Human Behavior"

AutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

So autonomous agents are real... This begs the question: Just tell the agent what the goal is and then it will manage itself forever?

The answer is yes.

You only need to set a goal for it, and the autonomous agent will complete the rest. It's like a really good employee or teammate. However, if you wish, you can design your own autonomous agents to contact you at certain key decision-making moments so that you can direct their work collaboratively in real time.

"This is primitive AGI. It is worth noting that by simply wrapping an LLM in a loop, you can get an autonomous agent that can reason, plan, think, remember and learn on its own. With the right packaging and prompts, LLM can unleash endless potential and flexibility. Although the whole concept is less than a month old, I can't wait to see how complex agents built from LLM will impact the world." ——Chen Siqi, founder and CEO of Runway

In addition to analyzing goals and assigning tasks, autonomous agents also have a series of capabilities, such as: #Browse the Internet and use applications;

  • Long-term and short-term memory;
  • Control your computer;
  • Use large language models (LLM) such as GPT to analyze and summarize, put forward opinions and give answers.
  • Additionally, these autonomous agents will come in all shapes and sizes. Some will run behind the scenes without the user knowing what they are doing, while some will be visible, as in the example above, allowing users to track the AI’s every “thought.”

"Autonomous agents will allow everyone to live like a head of state! Just ask and the autonomous agents will take care of the rest. You don't need to waste your brainpower on routine or mundane things ." ——Chris Yeh, one of the authors of "Blitzscaling" Summary of the latest news on a topic, say news about Twitter:

We tell the agent “Your goal is to find out the latest news about Twitter and send me a summary ";

Therefore, after the agent sees the target task, it uses OpenAI's GPT-4 and other AI to understand the content it is reading and proposes the first task, which is " Task: Search for Twitter-related news on Google";
  • The agent then searches for Twitter news on Google, finds popular articles, and returns a list of links. The first task is completed;
  • The agent then reviews its main goal (find the latest news about Twitter, then send a summary) and what it just accomplished (get a bunch of news about Twitter news link) and decides what its next task needs to be;
  • After that, it proposes two new tasks: 1) Write a news summary. 2) Read the content of a news link found via Google;
  • Now the agent pauses for a moment before continuing its work, it needs to make sure that the order of these tasks is correct. Are you sure you should write the abstract first? The agent disagrees and decides that the first priority is to read the content of the news link found through Google;
  • The agent reads the content from the article and then goes back to the to-do list again. It wants to add a new task to summarize the content, but the task is already in the to-do list, so the autonomous agent does not add it;
  • The agent checks the to-do list, only The remaining item was to summarize what it read, so it did that. At this point it will send the summary as requested by the user.
  • The diagram below shows how an autonomous agent works:

Source: BabyAGI by Yojei Nkajima

AutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

This new paradigm has just begun, but it is not perfect and has not taken the world by storm, but the concept is very powerful and is growing as With continuous development and experimentation, it will soon be integrated into our daily lives. "This will soon change many industries. By using autonomous agents, people can do many things more easily at the same time. Just give it a task and it will do it. So far, This is a very powerful concept... "——Barsee, founder of The AI ​​Valley Newsletter

After understanding what autonomous agents are to a higher degree, let’s discuss why autonomous agents Can it bring such a big opportunity?

Let’s dig into it.

“If we can get the information we need faster, will this free us up to focus on thinking and doing? Since this AI agent can perform more tasks, people can By investing less time in those tedious tasks, will people come up with better and more creative ideas?" - Marina Pérez, Director of Account Management at Octane AI

AutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

Why autonomous agents represent such a big opportunity

It’s clear that soon you’ll have the option of not only hiring humans as employees, but also in the form of autonomous agents AI.

"It won't be long before I believe you will see a huge increase in the number of 1-2 person startups using tools like AutoGPT and ChatGPT together. They will be able to achieve the 100's of benefits you previously had. The kind of progress that human-scale startups can expect. In the long term, I believe most jobs can and will be replaced by AutoGPT.” —Lore founder Nathan Lands

And they won’t As expensive as hiring humans are, they never sleep, never quit, and work incredibly efficiently.

"Part of my vision when I started Product Hunt in 2013 was the belief that the barriers to building software products would continue to fall, enabling smaller teams (or single individuals) to go faster than ever before. to create software independently. Driven by artificial intelligence and autonomous agents, this has never been more true than today. This has brought anxiety to some people and opportunity to others, who are taking advantage of this Technology scales to realize their ideas with fewer people and money. Ultimately, consumers will benefit through the continued fierce competition between companies and the continuous emergence of new solutions." ——Ryan Hoover, founder of Weekend Fund and ProductHunt

These autonomous agents will exist in every industry and be used for every task imaginable. The image below shows just a few examples:

AutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

The list could go on and on. What humans can do, autonomous agents will (eventually, but soon, and in some cases already) be able to do it better.

# "The music industry imposes too many unnecessary things between artists and success. These things cost artists nearly 35% of their net income. Autonomous Intelligence Entities will be able to develop and execute marketing strategies, interact with fans, build communities, book venues and negotiate contracts, etc. Saving artists money and time." ——Troy Carter, co-founder of Venice Music and former manager of Lady Gaga

How to seize the opportunity? There are two very real opportunities.

Create autonomous agents yourself and make them hireable for others;

Hire autonomous agents who can now help improve your Personal quality of life or business productivity.

“Autonomous agents are the next wave — not just in technology, but in business as a whole. Within 10 years, I predict there will be several multi-billion dollar companies built entirely from Autonomous agents operate. It’s inevitable.” — Ben Parr, co-founder and president of Octane AI

Imagine a world where one person built a company , whose team has only autonomous agents. In your lifetime, you'll likely see a team of one person be able to do this and achieve over $1 billion in market capitalization, which typically requires a very large number of people working together to accomplish.

"Personalization at scale is going to be a very interesting use case. You will be able to independently control the self-driving multi-step processes that humans perform today, including generating personalization Images, videos, websites, even emails and even mass calls.One of the use cases that can generate a lot of interest is sales development.」——Omar Pera, Head of Artificial Intelligence Product at Meta

Now, in the early stages, first movers, whether building autonomous agents or using them, will gain a huge advantage over competitors who have yet to take advantage of these systems.

"In the near future, I'd like to see lunch meetings, phone calls, and interviews appear on my calendar without me being involved in making them. Both my agents and their agents can do it At this point, all the details are taken care of. I just have to show up." - Hugh Howey, New York Times bestselling author of "WOOL"

By reading this article, You will already be ahead of 99% of the world. Let’s dive into more details about how these autonomous agents work.

"Autonomous agents have the potential to enhance the output of smaller content creators and community members, especially those with creative imaginations. This will be a boon to many Web3 projects. (" "Web3.0" is a derivative of the "World Wide Web", the underlying protocol of the current Internet. It means that machines can read any information, and websites can provide intelligent selection and better information (artificial intelligence) based on the information. The Internet has no Ubiquitous (Internet of Things), and more importantly, the ownership of data on the Internet will be decentralized.)" - Jeffrey Zirlin, co-founder of Axie Infinity

AutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

How autonomous agents work

You already have a general understanding of how autonomous agents work, but I think I will provide you with an overall framework version and gradually break down several autonomous agents. Agent examples can be helpful.

# "I now look at AI as a whole, and we are in the building stages of evolving it into an artificial intelligence assistant, like we see in the movies Like Jarvis in Iron Man or TARS in Interstellar.

Now is the time to build the framework. Because AI itself is still improving , the answers given may not be perfect, and may contain errors. But looking back at how much artificial intelligence has progressed in the past 6 months, I think we can hardly imagine the progress of AI in the next 1-2 years. So this is about early, Experiment quickly and plan for the future.” — Jenny Reece, Consumer Insights Manager at Microsoft

The following is a general framework for autonomous agents:

##Initialization goal: Define the AI's goal;
  • Task creation: The AI ​​checks its memory for the most recently completed X tasks (if any), and then uses it The goals and the environment of recently completed tasks are used to generate a new task list;
  • Task execution: AI executes tasks autonomously;
  • Memory storage: Tasks and execution results are stored in the vector database;
  • Feedback collection: AI collects feedback on completed tasks in the form of external data or AI internal dialogue. The results of this feedback will be used to inform the next iteration of the adaptive process loop;
  • New task generation: Artificial intelligence generates new tasks based on collected feedback and internal dialogue;
  • Task Prioritization: AI reprioritizes the task list by reviewing goals and looking at the last completed tasks;
  • Task Selection: AI prioritizes tasks from Select the top tasks in the list and continue executing them as described in step 3;
  • Iteration: The AI ​​repeats steps 4 to 8 in a continuous loop, allowing the system to information, feedback and changing needs. Gabriel Menezes, Director of Engineering at Octane AI People who automate jobs or repetitive tasks, I see that these agents have the potential to revolutionize the way we work, allowing us to redirect our mental energy toward more meaningful pursuits."
  • Example Display

Example 1: Social Media Manager Autonomous Agent

Suppose you don’t want to hire a social media manager to manage your social media accounts, but instead want autonomous agents to do everything for you at very low cost and with 24/7 intelligence.

"This is not just a virtual assistant. This is a revolution that accelerates all online work, research and even entertainment. Things that previously took hours, days, months to do online, Now it can be done in the background in minutes.” — Sharon Zhou, CS faculty at Stanford University and former machine learning product manager at Google

Here’s what an autonomous agent framework might look like:

  • Initialize Goals: Set initial parameters such as target audience, social media platforms, content categories and posting frequency;
  • Data Collection: Collect information about past social media Data on posts, user interactions, and platform-specific trends. This may include likes, shares, comments, and other engagement metrics;
  • Content Analysis: Analyze collected data to identify patterns, trending topics, hashtags that are relevant to your target audience and influencers. This step may involve natural language processing and machine learning techniques to understand the content and its context;
  • Content Creation: Based on analysis, generate content ideas and create social media that is appropriate for the platform and audience preferences post. This may involve using AI to generate text, images or videos, as well as incorporating user-generated content or curated content from other sources;
  • Schedule planning: based on platform-specific trends, audience activity and desired frequency determine the best time to post each piece of content. Schedule posts accordingly;
  • Performance Monitoring: Track the performance of each post based on engagement metrics such as likes, shares, comments, and click-through rate. If possible, collect user feedback to further refine your understanding of audience preferences;
  • Iterate and improve: Analyze performance data and user feedback to identify areas for improvement. Update content strategy, creation, and planning processes to incorporate these insights. Repeat steps 2-7 to continually refine your social media management system and increase its effectiveness over time.

"People will have personal agents that communicate with other people as well as corporate-owned agents. Most computing devices will be used primarily as communication devices to talk to agents." — —Conner Ruhl, Senior Software Engineer, Stability AI evolve due to the environment. This will help increase the engagement, reach, and overall effectiveness of your social media efforts.

"Another use case for autonomous agents that excites me is its application in music creation. By leveraging the power of AI-driven algorithms, these agents can analyze my personal preferences, favorites, etc. Genres, or even specific musical elements that resonate with me. They can then generate original melodies, harmonies, and rhythms, effectively co-creating music with me. This creative collaboration has the potential to broaden my musical horizons and allow me Being able to explore new styles and genres that I may not have considered before. Additionally, autonomous agents can provide valuable feedback on my compositions and provide suggestions for improvements, nurturing my growth as a musician. Artificial Intelligence and Human Creativity in the Music Creation Process The fusion in can lead to innovative and unique results, expanding the boundaries of artistic expression." ——Katya Sapozhnina, Product Director of Octane AI

AutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

##Example 2: Autonomous Agents Able to Run for Office

If you are running for an office and want to leverage an AI assistant for help.

I hope that the work done by the agent does not have to be difficult, but that it takes some time and effort. Things like booking a flight, I would happily outsource to an agent. 》——Sahil Lavingia, founder and CEO of Gumroad

  • Initial Goal: Win an election by securing a majority of votes;
  • Data Collection: Gather information about voters, demographics, key issues, campaign information, and other relevant Informational data;
  • Contextual analysis: Analyze collected data to identify trends, opportunities, and challenges. Use this analysis to refine initial goals into specific sub-goals, such as targeting undecided voters, increasing voter turnout in key areas, or improving campaign messaging on specific issues;
  • Task Generation : Generate tasks related to refined sub-goals, such as planning voter outreach, creating targeted ads, or developing policy recommendations;
  • Task Execution: Execute the highest priority tasks , allocate resources and assign team members as needed;
  • Performance Monitoring: Evaluate the effectiveness of completed tasks by tracking key performance indicators such as voter participation, public opinion, and fundraising metrics. Evaluate the success of individual tasks and the overall campaign's progress toward subgoals and initial goals;
  • Iterate and improve: Analyze performance data to identify areas for improvement and incorporate these insights to update the activity strategy. Repeat steps 2-8 to continually refine your campaign management system and increase its effectiveness over time.

"I'm very excited about the recursive self-cloning capability. AI agents can create copies of themselves, pass on task instructions, and start talking to their own siblings to get work done. This is A very remarkable but bizarre emergent ability.” — Jim Fan, NVIDIA artificial intelligence scientist D advantage, but imagine what it would be like once each campaigner had one... or more agents.

"I don't think everyone will use autonomous agents. Despite their ubiquity, as artificial intelligence develops, there will be a renaissance of human-involved work. Many people will rethink pens and Paper, want art made by humans... We will see many products and creations advertised as "entirely made by humans". It should become a very popular label soon. The faster the technology develops, the more I enjoy it Those who are completely offline for a long time will soon enjoy the time "off AI"." - Loic Le Meur, founder and CEO of PAWA | Loic

AutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

##Example 3: Autonomous Agents Teaching Mathematics

There will also be autonomous agents designed to teach children mathematics. .

"This is a groundbreaking paradigm with a lot of room for exploration. Although early experiments have limited agent search queries, we will see a wide range of research and support projects to equip autonomous agents. New tools. And each set of tools will significantly expand its potential use cases.” —Pete Huang, founder of The Neuron Daily AI newsletter

  • Initial goal: Determine children’s current math skill level and set up personalized learning paths to help them improve;
  • Data collection: through assessment, interaction and feedback to collect information about the child's learning style, learning process, and academic performance;
  • Contextual analysis: Analyze the collected data to identify the child's strengths, weaknesses, learning preferences, and All external factors that affect children's progress;
  • Task generation: Generate tutoring tasks according to children's needs and learning paths, such as selecting appropriate exercises, providing explanations, or providing real-life examples Examples and Applications;
  • Task Prioritization: Rank tutoring tasks based on their potential impact on children’s learning and skill development, and find a balance between challenge and engagement;
  • Task Execution: Execute the highest priority tasks, adjusting tutoring methods and content delivery as needed to maximize children’s learning mastery and engagement;
  • Performance Monitoring: Evaluate the effectiveness of tutoring by tracking key performance indicators (KPIs), such as progress toward learning goals, improvement in math skills, and child engagement and satisfaction;
  • Feedback Loop: Continuously monitor your child’s performance and update contextual analysis, task generation, and task prioritization steps based on new data and insights. Adjust initial goals and learning paths as needed to better support your child's math skill development;
  • Iterate and improve: Analyze your child's performance and update contextual analysis based on new data and insights, Task generation and task prioritization steps. Adjust initial goals and learning paths as needed to better support your child's math skill development. Repeat steps 2-9 to continuously improve the education management system and increase its effectiveness over time; Math tutors adaptively help and guide the process of a child's learning experience, with a focus on continuously improving and providing personalized guidance based on the child's needs and progress.

The future of autonomous intelligence

AutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

Humanity is now in the initial stages of developing autonomous intelligence. We poke around, break things, experiment, create things that are good or bad.

"By asking for the help of autonomous agents, they will turn your ideas into reality. These agents can act as friends, colleagues and collaborators, providing you with ample leisure time. I I really want to know, how will you choose to spend this newfound free time?” ——Kazuki Nakayashiki, co-founder and CEO of Glasp

Now, there are almost no commercialized autonomous agent products released, and such products are still in the development stage. But soon, that will change. Autonomous agents will start appearing everywhere.

“Instead of focusing on replacing people’s jobs, focus on empowering them. Making something “smart” used to mean making its data available through an API. The next generation of intelligence would be to ask how the product can better help you. For example, a "smart" email address might be able to act in interesting ways based on your preferences. If you're a shopping junkie, maybe it monitors emails to learn how you feel about it. Find out when the items you are interested in will go on sale, conduct price comparisons, and even negotiate prices on your behalf to privately understand your valuation of the item and the price you are willing to pay for it." ——Matt

, managing partner of Factorial Capital and investor in HuggingFace People will enhance their activities, decisions and actions through various autonomous agents. If we had neural implants at some point in the future, this would all happen naturally, just like thinking in your own head today.

"Everyone can have access to virtual researchers, assistants, writers or staff for free or at a small cost. This access is universal." ——Jeremiah Owyang, Artificial Intelligence Investor

The following is my prediction for the future of autonomous intelligence:

  • Multiple commercialized autonomous agents for gaming, personal use, marketing and sales will appear in 2023;
  • Various categories of businesses will appear in 2024 ubiquitous adoption of autonomous agents in every category imaginable by 2025;
  • In 2026, most people in first world countries will live their daily lives with the assistance of a large number of autonomous intelligence agents;
  • In the next 2-5 years, Most people will work for autonomous agents rather than humans.
  • "I see Holodecks using augmented reality, almost entirely driven by AI, with a lot of things happening with automatic and manual prompts. Yes, people will be AI jobs. Everyone uses them, but only a few know what they are or how to make them. The world is about to change profoundly due to the continued emergence of large language models (LLMs) and upcoming autonomous agents and systems.
  • LLM is the most universally accessible power that humanity has ever invented. Why? LLM can now be run on cheap computers without needing to be connected to a central server. That little engine contains essentially all of human knowledge. It’s mind-boggling , you can run it on devices that are not connected to the internet. Autonomous agents just make this holodeck run nearly autonomously. Everything from the weather to the pizza delivery basically happens automatically, with very little intervention from human input.” — —Robert Scoble, Chief Strategy Officer, Infinite Retina AI-First

The future is going to be crazy. So how do you build and use autonomous agents?

"In this future, everyone is likely to use autonomous agents in some capacity, whether for personal productivity, business operations, or creative activities. In most cases, people will act as these "Masters" of AI agents, setting goals for them and driving them forward. We will also "work for" AI agents, just as we must work within the constraints of companies, processes and other systems. However, I think AI agents will in many cases do better than the companies and systems in society today, and will create opportunities that benefit everyone." —— Joe Heitzeberg, co-founder of Crowd Cow

How to Build and Use Autonomous AgentsAutoGPT star量破10万,这是首篇系统介绍自主智能体的文章

You are now ready to dive headfirst into the world of autonomous agents. I'll list the resources you need to get started building or using autonomous agent agents.

"Find a specific B2B use case that contains a large number of repetitive tasks. Such as sales operations, advertising operations, project operations, accounting services, etc. There are many tasks to choose from now." —— Hustlefund co-founder Elizabeth Yin

“First, narrow the scope of your use cases as much as possible. Then, design a product that includes human-in-the-loop, And a way to evaluate the success of the process, and gradually increase the degree of automation, and finally expand to adjacent use cases." ——Itamar Friedman, co-founder and CEO of Codium AI

Building an Autonomous Agent

There are several different options for building an autonomous agent.

  • Build it yourself: Check out the framework I provided earlier and get started on the journey of building everything from scratch! This isn't as scary as it sounds. Recommended software solutions include OpenAI’s GPT-4, Pinecone vector database, and LangChain’s framework.
  • Auto-GPT: This is a popular open source option created by Toran Richards. It includes options such as connecting to the internet, using apps, long-term and short-term memory, and more.
  • BabyAGI: Another popular open source option, created by Yohei Nakajima. Although this one isn't connected to the internet yet, it's very concise at less than 200 lines of code.
  • Jarvis by Microsoft: Very similar to Auto-GPT and BabyAGI, but more powerful, provided by Microsoft and HuggingFace.

"I think we will initially have vertical domain autonomous agents. These agents are fine-tuned to a specific set of data so that they can function in that domain. So far, (The only?) two areas where we see heavy use of large language models (LLMs) are copywriting and programming. Extrapolating further, it makes sense to think that the AI ​​adopted in these two areas will start to become more autonomous. One way that may gradually emerge in the near future is that artificial intelligence will replace people giving prompts to trigger copywriting or coding, and they will automatically give you new suggestions for you to consider every day without requiring you to activate them first or Give them prompt." -Lonis Hamaili, creator of godmode.space

Using autonomous agents

Ready to have your own agent? Here are some options.

You can create your own agent as described above by selecting any of the options!

  • AgentGPT: Create and run autonomous agents (AutoGPT) from the website, no login required.
  • HyperWrite Assistant: Adds a chrome extension that allows you to issue commands to the browser, which then execute them.

"People from all walks of life can benefit from expertise and efficient methods previously reserved for society's elite. This kind of personal assistant The universalization of AI can lead to higher productivity and a more balanced work-life experience, allowing people to focus more on their interests, creativity and personal growth, while their AI assistants handle the more complex aspects of their daily lives. The monotonous part.” —Matt Shumer, founder and CEO of HyperWrite

Whether you know how to program or not, I encourage you to spend a few hours trying these things. It's not as complicated or difficult as it seems, and the quicker you get started, the quicker you'll understand autonomous agents.

“As an investor, I’m very excited about using autonomous agents to do the jobs of analysts and assistants, or at least to greatly enhance their jobs. They can Programmatically find deals under specific conditions, analyze specific factors, and then send me a custom email to start a conversation.” — Brayton Williams, Co-Founder of Boost VC

Autonomous agents can now be open to interpretation and innovation. With 99% of use cases yet to be created or tried, the possibilities are endless and the opportunity is in your hands.

"I'm very interested in the orchestration and modularization of smaller programming tasks to achieve larger end goals. We know that large language models are good at problem-based programming, But we haven't seen evidence that they can port an entire code base from Android to iOS, or even create an app from scratch. I think agents with the right orchestration scheme and memory structure may be able to achieve this goal." —— Curai Co-Founder and CEO Neal Khosla

The above is the detailed content of AutoGPT has exceeded 100,000 stars. This is the first article to systematically introduce autonomous agents.. For more information, please follow other related articles on the PHP Chinese website!

Related labels:
source:51cto.com
Statement of this Website
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn
Popular Tutorials
More>
Latest Downloads
More>
Web Effects
Website Source Code
Website Materials
Front End Template
About us Disclaimer Sitemap
php.cn:Public welfare online PHP training,Help PHP learners grow quickly!