LLMs take info. from internet and with training, can generate new information. Generative and Predictive AI have differences.
February 1, 2024 at 5:30 AM
Visible to public
Explain what Large Language Models are
Chatbots use Large Language Models is what Chatbots are using. They take the largest amount of information possible like everything available on the Internet. Uses training to be able to generate completely new information like to write essays or poems, have conversations, or write code. Companies are using them in all kinds of new ways. In education, teachers could use to generate lessons or tests. Students can use to make presentations like pictures, videos, text, etc. ChatGPT trains on all the information it can find on internet. Instead of learning from just the small amount of information, it looks at more such as tokens and code. It needs a lot of human tuning to get reasonably accurate results. It can often get things wrong. Is there actual intelligence? LLM produces amazing results like to create apps, websites, producing movies, and discovering new drugs.
Describe the difference between generative and predictive AI.
Generative AI can create images or videos and needs user prompts. Predictive AI uses probabilities based on previous texts. They are based on math from statistics. Needs to be trained to look at more, use a neural network. They are trained on a large body of information. Can learn to take in new information and make some answers based on probabilities. Predictive AI is like a brain's neural network that gets trained to look at a whole body of information. It needs humans to set it up. Once enough information is input, the neural network can learn from the information. It uses probabilities and training and give answers.
Karen Young