Prerequisite: None
The latest development in the advanced analytics arena focuses on ChatGPT. When this sophisticated new technology was introduced this past November by OpenAI, it triggered a huge interest in the business world and general public due to its ability to interactively automate generation of a wide range of textual outputs, including everything from magazine articles and marketing copy to program code builds, term papers, short stories, song lyrics, and on and on.
In the first five days after its release, more than a million people worldwide signed up to try out ChatGPT for free. What powers ChatGPT is a rapidly evolving approach called large language models. Essentially, LLMs—such as the GPT 3.5 model under the covers of ChatGPT—operate a bit like autocomplete engines on steroids. When trained on a vast corpus of pre-existing textual material, LLMs can intelligently respond to user’s textual prompts by generating content of surprisingly high quality. Many LLM-driven chatbots, such as ChatGPT, primarily generate textual outputs, while others can even generate realistic images, composite drawings, and other media outputs.
Some key use cases for LLM include:
- Conversational AI
- Intelligent question answering
- Automated text and image generation, completion, synthesis, classification, and summarization
- Machine translation
- Speech-to-text
- Sentiment analysis
- Sentence completion
- Word-sense disambiguation
In this talk, James Kobielus, TDWI’s senior research director on data management, will discuss the value that ChatGPT and other LLM-powered apps can unlock as well as the concerns that this technology has unleashed, such as its potential to generate factually inaccurate or undesirable content, propagate toxic content, or facilitate fraud. Jim will briefly demonstrate ChatGPT and will provide guidance for enterprises seeking to retool their data and analytics governance practices for the coming age of ubiquitous LLMs.
You Will Learn
- What is ChatGPT?
- What are the use cases for ChatGPT?
- How does ChatGPT use large language models?
- What are large language models generally?
- How do large language models work?
- How do you build, train, deploy, and optimize large language models?
- What are the leading large language models and who provides them?
- What skills, tools, and platforms do enterprises need to develop, deploy, manage, and govern large language models?
- What are the opportunities for large language models in data and analytics?
- What are the challenges involved in building, training, optimizing, and governing large language models?
- What are the business risks of ChatGPT and other large language models?
- What new enterprise analytics governance practices are necessary to mitigate these risks?