AINoon Lesson 1

Get Ready for AINoon!

Thanks

  • To the host for the great venue!
  • To our sponsors

Administrivia

  • Fire escapes
  • Toilets
  • Cleaning up after ourselves
  • Wi-Fi

Welcome to AINoon!

Hands up if you’ve ever:

  • Used Google’s “AI Overview” instead of opening search results?
  • Asked ChatGPT a question?
  • Used AI to automate a task?
  • Used AI to build an app or write code?

What we’ll cover in AINoon

  • Leading applications of Generative AI
    - Spot opportunities to increase productivity
  • Hands-on practice in follow-along tutorials
  • Demystifying how Generative AI works
    - Understand strengths and limitations
  • Risks and issues to consider when using AI
  • Providing a forum for questions and assistance

How to get the most out of AINoon

  • As much as possible, don’t do emails and work
  • Follow along with the tutorials
    • We have a mix of experience levels
    • If you’ve used a tool before, consider the results in the context of ideas we’re discussing
    • You might still learn a new tip or trick!

AINoon Structure

  • Lesson 1:
    • Presentation: Terminology and business use-cases
    • Tutorial: Chatbots for business
  • Lesson 2:
    • Presentation: How AI works
    • Tutorial: Build a chatbot on your documents with Chatbase
  • Lesson 3:
    • Presentation: Common patterns: RAG, tools, and agents
    • Tutorial: Build an agent with Zapier
  • Lesson 4:
    • Presentation: AI risks and challenges
    • Tutorial: “Vibe-coding” an app with Gemini

All slides, tutorials, and optional homework on technoon.org/ainoon

Questions?

Intro to Generative AI

  • Establishing a common vocabulary
  • Who’s who in the zoo: companies and services
  • How generative AI is being used by businesses

AI Terminology

Artificial Intelligence (AI) ~1950s

  • General term for computers making “intelligent” decisions
  • E.g. Hand-crafted programs that can play checkers

Machine Learning (ML) ~1980s

  • Approach to AI where computers “learn” from patterns in data
  • E.g. Learning from many past emails to identify spam

Deep Learning (DL) ~2000s

  • Approach to ML based on very large (artificial) “neural networks”
  • E.g. Recognising objects in images, text translation

Generative AI (GenAI) ~2020s

  • Application of DL to generate text, images, audio, video, etc.
  • What most people mean by “AI” these days
  • The focus of this course

Source: blogs.nvidia.com/blog/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/

Chatbots are the most prominent application of GenAI

  • ChatGPT (OpenAI)
  • Claude (Anthropic)
  • Gemini (Google)
  • Microsoft Copilot (based on OpenAI tech)
  • Grok (X)
  • DeepSeek

What can chatbots do for business?

  • Drafting emails and documents
  • Summarising documents and meetings
  • Brainstorming a wide variety of ideas
  • Generating code to assist programmers
  • Data extraction and transformation
    • Turn unstructured text and images into structured data
    • E.g. Identify which team an IT ticket should go to
  • Custom agents: chatbots that retrieve info and take action
    • Answering questions from organisation knowledge bases
    • Personalised customer service
    • Automated handling of events - e.g. IT tickets

Using chatbots for business

  • Public chatbots train future chatbots with your data
  • Companies deploy internal chatbots to protect data
    • Often using business-to-business service offerings
    • E.g. From Microsoft, Amazon, or Google clouds
  • You can run open chatbots on your own computers

Tutorial Objectives

  1. Using a chatbot for business use cases:
    • Brainstorming and drafting
    • Automating web searches
    • Summarising documents
  2. Common accuracy issues to be aware of
  3. Tips for effective prompting

Tips for crafting better prompts

  • Each prompt should request one thing
  • Start prompts with a persona
    - E.g. “You are an expert in project management…”
  • Ask it to “explain step-by-step”
    - Can get better answers to more complex questions
  • When it’s not doing what you want, be more specific
    - Though it won’t always follow instructions exactly!
  • Seek more tips online for specific use-cases
    - It’s all art/craft, not science/engineering
    - Advice will likely change as chatbots change

The Golden Rule of AI

Don’t trust the output of an AI unless you can verify it

  • Bad: Summarising a document you haven’t read
  • Good: Summarising a document you have read
  • But: For some use cases, speed may be more important than accuracy

Homework

Remember: Don’t use private data with public services!

  • Generate a video with Google’s Flow
    • Look at the gallery on Midjourney - note the detail in the prompts
  • Practice prompting
    • Describe a photo with text to prompt ChatGPT to replicate it
    • Keep refining your prompt to more closely match the photo
  • Generate speech from text with ElevenLabs