In what some describe as one of the most important advances in the field of generative artificial intelligence and a possible dethroner of Google Search's 20 year + monopoly, AI research and deployment company OpenAI has launched its "ChatGPT" conversational AI tool.
According to OpenAI, dialogue format makes it possible for ChatGPT to answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.
ChatGPT is a sibling model to InstructGPT, which is trained to follow an instruction in a prompt and provide a detailed response.Trying out ChatGPT is absolutely free and you can do it here and get your mind blown away.
How ChatGPT Works
 |
| How ChatGPT works |
OpenAI trained the model using Reinforcement Learning from Human Feedback (RLHF), using the same methods as InstructGPT, but with slight differences in the data collection setup. They trained an initial model using supervised fine-tuning: human AI trainers provided conversations in which they played both sides—the user and an AI assistant. They gave the trainers access to model-written suggestions to help them compose their responses.
To create a reward model for reinforcement learning, they needed to collect comparison data, which consisted of two or more model responses ranked by quality. To collect this data, they took conversations that AI trainers had with the chatbot. They then randomly selected a model-written message, sampled several alternative completions, and had AI trainers rank them. Using these reward models, they can fine-tune the model using Proximal Policy Optimization. They performed several iterations of this process.
ChatGPT's Limitations
OpenAI has disclaimed that despite its brilliance, the tool hjas its own limitations which they have outlined below:
- ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers. Fixing this issue is challenging, as: (1) during RL training, there’s currently no source of truth; (2) training the model to be more cautious causes it to decline questions that it can answer correctly; and (3) supervised training misleads the model because the ideal answer depends on what the model knows, rather than what the human demonstrator knows.
- ChatGPT is sensitive to tweaks to the input phrasing or attempting the same prompt multiple times. For example, given one phrasing of a question, the model can claim to not know the answer, but given a slight rephrase, can answer correctly.
- The model is often excessively verbose and overuses certain phrases, such as restating that it’s a language model trained by OpenAI. These issues arise from biases in the training data (trainers prefer longer answers that look more comprehensive) and well-known over-optimization issues.12
- Ideally, the model would ask clarifying questions when the user provided an ambiguous query. Instead, our current models usually guess what the user intended.
- While we’ve made efforts to make the model refuse inappropriate requests, it will sometimes respond to harmful instructions or exhibit biased behavior. We’re using the Moderation API to warn or block certain types of unsafe content, but we expect it to have some false negatives and positives for now. We’re eager to collect user feedback to aid our ongoing work to improve this system.