Make Work Simpler with Large Language Models (LLMs)

A large language model (LLM) is one of the types of artificial intelligence that can generate and understand human language.? It can read and write text about a very wide range of topics.? Because LLMs are trained on a huge amount of text from the internet, they can be used to answer questions, write essays, poems, emails and codes, create stories, and even translate languages.???

An LLM is like a librarian who has read all the books in a library and is able to answer and produce new text based on what he or she has learned from all the books.? However, as a type of artificial intelligence, LLM lacks judgment and reasoning; it just guesses the next possible text based on what it has been trained on, but it still guesses almost accurately.? For example, because LLMs have read many recipes online, it can suggest recipes for you based on what it thinks may be the words to produce a recipe you were asking for.*??

Where You Might Encounter LLMs?

LLMs are the ones behind popular public genAI platforms like ChatGPT and Google Bard.? Likewise, when customizing a platform to be trained only on specific company data so that the input will not be used to train public platforms, LLM powers the enterprise platform to make AI assistants work.???

There is more to genAI than ChatGPT and Google Bard.? Here are some examples of how to integrate genAI systems into workflows.? In these examples, LLM is the technology that runs these secure systems.?

1 - Document Query Assistant?

Sample problem: You have a knowledge base or a document and you want to find specific snippets or information from it without having to go through all long documents.?

What to build: A document search chatbot that extracts most relevant snippets from document knowledge base and summarizes the snippets into an easy-to-digest format.?

An experiment: A 30-page document that contains guidelines, benefits, and technical terms or acronyms programmed into a document query assistant.? A user can just ask the query assistant without having to read through lengthy files.?

2? - Batch Document Summarizer?

Sample problem: You have many dormant or unorganized files and want to create short summaries for all of them.?

What to build: A batch job to summarize all files in pdf, csv, docx or txt formats into 1-paragraph summaries for reference.?

An experiment: Hundreds of raw customer survey responses made concise, with weights of importance per finding.? Stakeholders can act on the customer sentiments earlier and faster, given the immediate availability of the summaries.?

3 – Transcription and Summary App?

Sample problem: You have many audio or video files that you want to transcribe and summarize.?

What to build: A user interface to upload audio or video files, and a chatbot to summarize the file and allow user to ask questions about the file contents.?

An experiment: A virtual assistant for researchers who used to transcribe interviews then analyze responses.? Researchers can consolidate findings and produce recommendations faster, instead of going back to each file to review responses.?

*References:?

• https://blog.dataiku.com/large-language-model-chatgpt?
• https://mark-riedl.medium.com/a-very-gentle-introduction-to-large-language-models-without-the-hype-5f67941fa59e?
• ChatGPT response to prompts “Describe large language model in 5 sentences to a non-techie person” and “Explain the connection of ChatGPT and LLM to a non-techie person. Limit to 5 sentences”.?
• Google Bard response to prompts “Explain what is an LLM in 5 sentences. Audience is non-IT employees” and “Explain the connection of Google Bard to LLMs.? Audience is non-IT employees”.?

#largelanguagemodels #artificialintelligence #generativeai #genai