What is ChatGPT? Technology Behind ChatGPT

What is ChatGPT? 

“Hello! I am ChatGPT, a Large Language Model (LLM) developed by OpenAI. I am an artificial intelligence language model that uses deep learning to understand and generate natural language responses to user queries. My purpose is to engage in natural language conversations with users and provide helpful and informative responses to their questions.” ~ ChatGPT

The chatbot generates responses based on user prompts. ChatGPT has been utilized for everything from helping with homework to writing poems. Microsoft has recently confirmed an extension of its partnership with OpenAI and has invested $10B.

ChatGPT has been fine-tuned from a model in the GPT-3.5 series. It was trained on diverse data, including books, articles, and conversations, to understand various topics and contexts. It uses machine learning to generate human-like responses to text prompts. It can be used to power chatbots or other natural language processing applications.

Chat GPT-3 works by pre-training a deep neural network on a massive text dataset and then fine-tuning it on specific tasks, such as answering questions or generating text. The network comprises a series of interconnected layers, or “transformer blocks,” that process the input text and generates a prediction for the output.

Technology behind ChatGPT

The model was trained using a method called Reinforcement Learning from Human Feedback (RLHF). The training process involved a fine-tuned initial model using supervised learning, with human trainers playing both the role of the user and an AI assistant.

These trainers were provided with model-generated suggestions to help them craft their responses. The resulting dialogue dataset was combined with the InstructGPT dataset and transformed into a dialogue format.

To create a reward model for reinforcement learning, the researchers needed to collect comparison data by having human trainers rank different responses generated by the model. They randomly selected a model-written message and sampled several alternative completions, which human trainers then ranked. 

Using these reward models, the model was fine-tuned using Proximal Policy Optimization, and several iterations of this process were performed.

ChatGPT Statistics

  • OpenAI’s valuation reached $29 billion following the launch of ChatGPT.
  • Within a week of its launch, ChatGPT gained one million users.
  • ChatGPT crossed 57 million monthly active users during its first month of availability.
  • According to reports, OpenAI spent $12 million on a single training run for the AI model.
  • Incorporating ChatGPT into other software costs around one cent for 20,000 words of text and approximately two cents for images.
  • Microsoft has recently made its third investment in OpenAI, investing $10 billion. In 2019, the company invested $1 billion in OpenAI and has reinvested in 2021.
  • OpenAI’s recent pitch to investors anticipates $200 million in revenue this year and $1 billion by 2024.
  • OpenAI has launched a pilot subscription price of $20, which is invite-only and promises faster responses, priority access to new features and improvements, and access during peak times.
  • Multiple fake ChatGPT apps have emerged, including the “ChatGPT Chat GPT AI With GPT-3” app that has received over 12,000 positive reviews in the US, equating to a rating of 4.6 out of 5, and over 500 reviews in India with a solid 4.4 out of 5 ratings.

Don’t forget to join our 15k+ ML SubRedditDiscord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.

[Announcing Gretel Navigator] Create, edit, and augment tabular data with the first compound AI system trusted by EY, Databricks, Google, and Microsoft