Have you heard of ChatGPT or large language models (LLMs), like the new Bing or Google’s Bard? ChatGPT has become one of the fastest-growing apps of all time, by utilizing LLMs as software algorithms trained on huge text datasets to understand and respond to human language in a very lifelike way. Although powerful, these applications could only go as far as completing one task at once, requiring human interaction (prompts) afterwards – which limited the complexity of their output. Enter Auto-GPT: a technology that attempted to overcome this limitation with its simple solution.
Auto-GPT was created by Toran Bruce Richards and is built on GPT-4 to construct sentences and prose based on text it has studied. It is open source, and utilizes Auto-GPT instances as independent project managers to work on small tasks before comping them into a larger, finished result. Auto-GPT takes it one step further by being able to go deep into the internet, and even having a better memory than ChatGPT – it can build and remember long chains of commands.
But what exactly can Auto-GPT do? Not only generating code and researching tasks, but this AI Agent monitors businesses and provides automatic recommendations on how they could be improved. Furthermore, it can develop software applications from start to finish; research new technologies and companies; look for the best golf clubs, and even attempt to come up with an AI solution to “destroy humanity” – although thankfully, its text output is still limited.
If it successfully reaches the “holy grail” of AI and becomes an AGI, or general, strong AI, this would be a large step forward from just using narrow AI. If the process of building these agents becomes more efficient, it could revolutionize the way we work, play, and communicate. As it has the ability to better itself, opinions of experts such as Nick Bostrom need to be taken into consideration, as this could create a new moral and ethical quandary – with the potential for bias and spreading of harmful content.
Toran Bruce Richars is a machine learning engineer and the creator of Auto-GPT. He’s worked on several projects and was inspired to create Auto-GPT because traditional AI models struggled to adapt to tasks that could not be adequately completed by ChatGPT, even after careful thought was put into it. His ambitious mission to revolutionize the way AI works has the potential to have a huge impact on the future.
