The Lightwave 

Practical Insights for Skeptics & Users Alike…in (Roughly) Two Minutes or Less

"Technology is just a tool. In terms of getting the kids working together and motivating them, the teacher is the most important" - Bill Gates

This week we looked at:

1. Deep Learning (Issue 7)

2. Neural Networks (Issue 8)

3. Large Language Models (Today!)

Powered by the aforementioned Deep Learning and Neural Networks, Large Language Models (LLMs) can understand* and generate human-like text and visuals.

Traditional Large Language Models were traditionally designed to focus on text, but these have quickly evolved to become multimodal: many LLMs (Like ChatGPT) can interact with images, video, and audio—and were trained with data sets to do so.

LLMs

For many people, Large Language Models are the lens through which they view AI…though it’s important to note that LLMs and Generative AI represent only a fraction of Artificial Intelligence.

In other words, Large Language Models are what many people fear: Will they take my job? Will they make me redundant?

These fears are well-founded: In 2025, an estimated 50% of digital work is expected to be automated.

What can LLMs Do?

Lots, but here are some key ones:

• Provide instant information: Like having a knowledgeable assistant available 24/7.

• Create content: From blog posts to social media updates.

• Offer customer support: Power chatbots that can handle complex questions.

• Analyze data: Quickly find important information in large amounts of text.

• Write basic computer code: Turn descriptions into working code.

The power lies in the prompt: The specific instruction or question given to an AI system like ChatGPT to guide its response or output.

Example:

Let’s say you have to schedule appointments across numerous calendars.

Perhaps you start with the following prompt:

I need to schedule a series of meetings with 5 team members and 3 external clients over the next month. Each meeting should last 1 hour. Team members are available Monday to Friday, 9 AM to 5 PM. Clients are only available Tuesdays and Thursdays, 10 AM to 3 PM. Please create a draft schedule that maximizes efficiency and minimizes conflicts. Then, generate personalized email templates to send to each participant for confirming their availability.

And the model might respond with something like this:

‘To maximize efficiency and minimize conflicts, we can schedule meetings as follows:

Week 1

• Tuesday

  • 10:00 AM - 11:00 AM: Meeting with Client 1

  • 11:30 AM - 12:30 PM: Meeting with Client 2

  • 1:00 PM - 2:00 PM: Meeting with Client 3

• Thursday

  • 10:00 AM - 11:00 AM: Meeting with Client 1

  • 11:30 AM - 12:30 PM: Meeting with Client 2

  • 1:00 PM - 2:00 PM: Meeting with Client 3…”

And so on.

For the last post of this week, I am going to focus on showcasing some interesting use cases I’ve come across in the past.

Then next week we’ll focus on prompting.

And PS: Happy Fourth of July!

For more, visit www.NorthLightAI.com