**optimising everyday workflows with ai** --- - setting context - how to think about ai - interactive demos - putting it into action - q&a --- **a few definitions** --- **generative ai** refers to artificial intelligence systems designed to create content, such as text, images, audio, and video, that mimic human-like outputs. it learns from vast datasets to generate new, original content that resembles the learned information, often used in applications like chatbots, art creation, and content personalisation --- **large language models** (llms) are ai technologies designed to process, understand, and generate text in a way that resembles human language. these models leverage massive datasets to learn language patterns, enabling sophisticated tasks like conversation, text completion, and question answering --- **why** ---  --- <img src="https://hackmd.io/_uploads/H1aUYMSq6.png" height="400" />  --- **to allow us to focus on the more enjoyable and creative parts of work** <img src="https://hackmd.io/_uploads/SkgpMo696.jpg" height="400" /> --- **why not** --- **feeling fomo because of all the hype around ai (it's probably overblown)**  --- **feeling left behind**  --- **on those who do not understand the hype**  --- **how to think about ai** --- **no instruction manual** *proficiency in working with ai comes through experimentation and experience* <img src="https://hackmd.io/_uploads/rkC8xQHqa.png" height="300" /> --- <img src="https://hackmd.io/_uploads/HJNTxUC9a.png" height="400" /> --- **centaur** <img src="https://hackmd.io/_uploads/B1XIDfrcp.jpg" height="500" /> --- **ai won't replace humans -- but humans with ai will replace humans without ai** --- **intern** - infinitely helpful intern who is a little naive and wants to make you so happy that they will make up facts rather than disappoint <img src="https://hackmd.io/_uploads/Hkm_acTcp.jpg" height="250" /> - it should come as no surprise that they'll hallucinate, lie and be over-confident --- in the context of large language models and ai, **"hallucination"** refers to the generation of false or misleading information that is not grounded in reality. it occurs when ai produces **confident but incorrect or fabricated** outputs during tasks like text generation or question answering --- **reasoning engine** - **it's not google** so you won't get good results if that's how you use it - chatgpt is **not a knowledge database**, it's a **reasoning engine** --- **open book tests** - when you ask it a question, give it all the context necessary to give you the answer - in short, **you get out what you put in** --- **a brief introduction to prompting** (but first a few more definitions...) --- **prompt**: a text input given to an ai model to generate a response or perform a task. prompts guide the ai in understanding what the user is asking for --- **context window**: this refers to the amount of text (measured in tokens) the model can look at in one instance to understand and generate responses --- **tokens**: units of text, such as words or characters, that serve as the input for language models. in ai, a token isn't just a word but can include punctuation and parts of words in general: **1 token = 0.75 words** --- # A guide to prompting MICHAEL TAYLOR --- "You’re communicating a prompt every time you brief your designer, explain how to do something in Excel to an intern, or give a presentation to your manager. Every manager is a prompt engineer, using their communication and data analysis skills to align the team toward a common set of goals. Even your employment contract is a prompt: a standardized template of language designed to align your behavior with the commercial goals of the organization." --- The four pillars of prompting* 1. Give direction: Describe the desired style in detail or reference a relevant persona. 3. Specify format: Define what rules to follow and establish the structure of the response. 3. Provide examples: Supply a diverse set of test cases where the task was done correctly. <!-- 4. Evaluate quality: Identify errors and rate responses, testing what drives performance. --> 5. Divide labor: Split tasks into multiple steps, chained together for complex goals. --- EXAMPLE Write a blog post on productivity with time blocking. --- <img src="https://hackmd.io/_uploads/HkpmYLR5p.jpg" height="550" /> --- **1. Give direction: Describe the desired style in detail, or reference a relevant persona.** <img src="https://hackmd.io/_uploads/rkr3YUAq6.jpg" height="550" /> --- <img src="https://hackmd.io/_uploads/B1bJc8AqT.jpg" height="550" /> --- **2. Specify format: Define what rules to follow and the required structure of the response.** <img src="https://hackmd.io/_uploads/BkjH9ICcp.jpg" height="200" /> --- <img src="https://hackmd.io/_uploads/B1JuqIAqp.jpg" height="550" /> --- **3. Provide examples: Supply a diverse set of test cases where the task was done correctly.** <img src="https://hackmd.io/_uploads/H14Cc8Rca.jpg" height="300" /> --- <img src="https://hackmd.io/_uploads/BkKls80cT.jpg" height="550" /> --- **4. Divide labor: Split tasks into multiple steps, chained together for complex goals.** <img src="https://hackmd.io/_uploads/B1u7jURcT.jpg" height="550" /> --- <img src="https://hackmd.io/_uploads/r11rjLR9a.jpg" height="550" /> --- **lay of the tooling landscape**  --- **workflow demos** --- **over to you** http://tinyurl.com/3557m35v