📨 Subscribe | 🤖 AI Training | 🚀 Courses

Today’s Thoughts ☠️

Ahoy there 👋,

This week found me delivering a workshop on how to wield the dark magic of AI prompting for L&D folks at the Learning and Development Institute in Ireland.

It was a great event.

I had ooh’s, ah’s and screams of terror when showcasing the power of AI. What more can this humble long haired, tea drinking nerd ask for? Not much.

I noticed a common thread once more when unpacking how to engage with conversational generative AI tools vs what we’ve been used to with search engines. A lot of people talk to ChatGPT, and others, like they do with a google search. When that doesn’t work, they give up at the first try and deem the tool useless.

Today, we’ll explore how to build the meta skill of AI prompting and why that requires a reset on what you know today.

Get your tea or beverage of choice ready, 🍵.

We've got lots to discuss!

→ Much love to today’s sponsor, You Can Now ❤️

In today’s chat:

• 👩‍💻 How To Learn The Meta-Skill Of AI Prompting For Quality Results

• 👀 My honest advice on starting a one person business: 1 year later Q&A

• 🤝 4 ways I can support you

THE BIG THOUGHT
How To Learn The Meta-Skill Of AI Prompting For Quality Results: Forget Everything You Know About Googling

It’s not your fault that you don’t know how to craft useful AI prompts to product quality outputs.

We’re still very early in this game.

In reality, no one showed you how the language of these tools works so you don’t get what you expect.

I coined this the fail fast fallacy.

I also like the term prompt and ghost because that happens too much. It’s like people are in physical pain if they’re dared asked to ask another question after the tool response. Imagine if that’s how we talked to each other.

You gotta know how to talk to machines. A lot of people don’t and they blame the machines.

It’s human error - plain and simple.

If you treat a conversational AI like a google search your results will always suck - fact.

Fret not, we can help you if you’re reading this right now thinking “Shit, this is me!”. A little bit of education, guidance and practice will change that.

I’m going to break this down into these sections:

1. How we’ve navigated the digital world for 25 years

2. How this is replicated in popular LMS and LXP platforms

3. The difference between a Google search term and AI prompts

4. How to unbundle yourself from the ‘google search mindset’ with AI

5. The universal AI prompt framework for critical L&D tasks

How we’ve navigated the digital world for 25(ish) years

Do you remember the time before Google?

I can. I was born in the late 80’s so my generation is probably one of, if not, the last to not come tumbling into this world surrounded by search engines, 24/7 connection and people posting dumb stuff for the world.

Some of you might be young enough to not know of this world.

While that makes me feel super old. You might find this nostalgic trip fun. I mean you missed the delights of AOL (America online) with it’s classic connection and browser.

Don’t get me started on the wonder that was “Ask Jeeves”. An early search engine moulded around a fictional english butler called Jeeves.

This was peak innovation and 12 year old me could not get enough of this virtual butler finding information for him. It could have been better at helping me amplify my Pokemon card obsession though.

Everything changed in late 1998 when a startup called Google hit the scene with it’s ‘Google search engine’. It had been in testing behind closed doors at Stanford University since 1996 but late 98 is when us lucky everyday folk got hold of this era-defining tech.

You don’t need me to tell you the rest because you know it.

No one says ‘I’ll search the internet”, they say “I’ll Google that”. The ultimate category king.

The thing is Google has been our companion to navigate the digital highways ever since. We do everything from research, looking at cute cat pics, best restaurants in x city, looking up old friends to cry ourselves to sleep over how successful they’re and we’re not (def don’t do that) or to find that one YouTube video to fix your bathroom light.

It’s remained the number one way to interact with the digital space, so by default, it is the number one language shared by each of us. That has worked well until now.

Before we get to that, let’s explore how this same tech lives in learning tech.

How this is replicated in popular LMS and LXP platforms

If keyword search has made Google a billion dollar company, why not use the same or similar tech for other content hosting systems?

That’s what a lot of tools have done for a long time now.

Most LMS platforms often use keyword search as their backbone content search infrastructure. I say ‘most’ because I don’t need you LMS providers who read this to be sending me hate mail - chill!.

Like Google, this help users find relevant courses and content. You enter your specific terms and the LMS will return results that contain those keywords in the course titles, descriptions, or content.

LXPs are a little different, or so they tell us.

Many use a combination of both keyword search, where search queries are unpacked into individual words and then matched within the content's title, tags, and descriptions, plus natural language processing (NLP for short).

Natural language processing

This is not going to turn into the last part of the Matrix 2 where Neo and the architect meet to explain what the Matrix is.

I’ve watched that scene about 10 times and still don’t understand what the matrix is.

Anyway, back to NLP. This enables users to enter search queries in natural language, using conversational phrases or questions instead of just keywords. NLP algorithms process these queries to understand the user's intent and context.

Dependent on how good the LXP is, a few things are available here.

Rather than just matching keywords, LXPs use semantic search which just means it can understand the meaning and context behind the query. This enables the platform to retrieve relevant content even if it doesn't contain the exact words used in the search.

As an example:

Let's say you search for "leadership skills for managers".

With traditional keyword-based search, the LXP would look for content that contains those exact words or phrases within titles, descriptions, or tags. However, with semantic search powered by NLP, the LXP can understand the intent and meaning behind the query.

So, it recognises that the user is looking for learning resources related to developing leadership competencies for people in managerial roles.

The NLP algorithms would break down the query into its core components:

• Intent: Learn about leadership skills

• Target audience: Managers

It would then search the content library not just for literal keyword matches, but for semantically related concepts. This could surface results like:

• "Effective Communication for Leaders" course

• "Motivating and Inspiring Teams" video series

• "Emotional Intelligence in the Workplace" ebook

• Discussion thread: "What makes a great manager?"

Even if those content items don't contain the exact phrase "leadership skills for managers", the semantic search can identify them as highly relevant based on the contextual meaning extracted from the query.

Pretty smart.

Another useful feature is sentiment analysis which is deployed to understand the emotional tone or sentiment behind search queries, which can further refine the search results.

This is the same tech that google uses too. It’s like search on steroids. The thing is its still pretty basic knowledge retrieval which you as the human must assess.

The difference between a Google search term and AI prompts

For the uninitiated, when I say AI in this article I’m specifically talking about the generative family of AI models.

There are many different types of AI in our world.

Don’t let all the snazzy marketing campaigns fool you. To work with conversational Gen AI tools like ChatGPT, Microsoft Co-pilot, Google Gemini and the rest we must use a prompt.

A prompt is just an instruction.

You send an instruction to the tools database which then runs that query. No need to overcomplicate it here. Prompts need to provide a detailed, context-rich instruction or question to guide the AI in generating a specific response or performing a task. Someone classed this as prompt engineering.

There’s no actual engineering involved, fyi. Unless you class crafting words as engineering.

Ernst Hemingway might disagree.

The effectiveness of any conversational AI's response heavily depends on how well the prompt is crafted. I cannot not overstate this. As of 2024, I’m sure this will be etched onto my grave.

If you want to get nerdier, you can deploy techniques like chaining prompts, where complex tasks are broken down into simpler prompts, and guided reasoning.

Like I said, it has a sort of language and art to it.

One that’s always evolving.

Google search terms

Using search terms on Google, on the other hand, involves entering keywords or phrases to find information that already exists on the internet.

The effectiveness of your search results depends on the choice of keywords and how well they match the content available online. So, if the marketer was smart enough to stuff their article with those key words, they win the search lottery.

Google's algorithms analyse these keywords to retrieve and rank web pages based on relevance, quality, and other factors.

Unlike prompting an AI, where the goal is to generate new content or responses based on the input, using search terms on Google is about finding existing information.

⏱️ In sum: Prompting an AI and using search terms on Google serve different purposes and involve different processes. While AI prompting is about generating new content based on structured input, using search terms is about finding existing information through keyword matching.

How to unbundle yourself from the ‘google search mindset’ with AI

As L&D pros (I assume most of you reading this are), this bit should excite you.

To use conversational AI tools better, we must re-shape our mindset on what we’ve done these last 25 years. Easy, right? In essence, we need to learn new skills and behaviours.

Those are the things we’re in this game to do.

Before we sink our teeth into the final part of our chat, let’s look at steps you can take to unbundle your mindset:

1. Recognise how we’ve interacted with the digital world to date

2. Get clear on the difference between that and work with generative AI tools

3. Know when to use each method for what tool

4. Learn how to write AI prompts for any generative AI tool with the framework below

You’re welcome.

The universal AI prompt framework for critical L&D tasks

Prompts for AI can come in many forms for different outcomes.

Looking for a quick answer to a simple question? Use a 1-2 sentence prompt. But want to unpack a complex work task with layers of actions? You need a different approach.

Whereas a one shot prompt like “What’s the weather in Ibiza on average in August?” is a simple string of words for any conversational AI tool to answer. Asking it to review, ideate and share how to create a learning strategy for your company is not so straightforward.

To get the best results from something like a ChatGPT, you need 3 things:

• Context

• Task outline

• Constraints

This is a universal approach you can use with any tool. Let’s unpack each of these:

Context

What does the LLM need to know to successfully support you?

Here’s some ideas:

• Your organisation

• Team

• Roles and work they do?

• Specifics on the task

• What have you done before?

• What is the role it’s playing? [If role-playing or coaching]

Task

• Outline the task

• What does success look like?

• What are the essential components of the task?

• Keep it clear and simple

• How should the output be structured? Bullets, sentences or paragraphs.

Constraints

• What should the LLM not focus on?

• What must it not consider?

• Should it only use its training data or connect to the internet or both?

• Should it only use the data you’ve provided?

If we put this all together, it can look something like this:

# Context

I'm crafting my organisations L&D strategy for the year ahead. We’re a scale-up business with 800 employees in 5 global offices. We’re limited with our resource and budget to deliver. Our goal is not to do everything, but do the top 2-3 things that matter most.

Our strategy for the last few years has become stale and not working towards what we want to achieve.

Top things on our employees minds include:

- Having the right skills for the role

- Learning from their peers

- Manager support and coaching


# Task

You will help me get clarity on how I can work with my team to build a relevant and meaningful strategy for our organisation.

To do this, you will ask me questions to help get clarity to build a better picture of where we can go.


# Constraints

Keep questions short and relevant. They should be quick fire rather than in-depth. Let’s keep questions to a minimum of 3 at a time.

You might notice I’m using ‘markdown’ structure in these prompts. This helps AI tools better understand the instructions you provide by using heading, bullets etc.

You can learn more about that in this video.

Final Thoughts

We’ve covered a lot of ground today.

It’s taken me 10 hours, 5 cups of tea and probably a few more hours of procrastination to put this together. I hope you find this as helpful as I enjoyed writing it.

• AI prompting is a new language you can learn

• Look at it as a universal meta skill for at least the next few years

• Use the write language/operating system for the right tech

If you’d like direct help from me on any of this, you can:

1. Take my prompt writing for business 1hr masterclass

2. Become the envy of your team in the AI For L&D Crash Course

3. Hire me to work with your team to craft an intelligent research based approach to AI at work

Till next time.

SMART THOUGHTS

🤔  My honest advice on starting a one person business: 1 year later Q&A

This month I celebrated one year as a one-person business.

I’ve been running STT for about 5 years, but 2023 was when I decided to make this my full-time focus. It’s been rewarding, challenging and a constant flow of learning.

I get asked lots of questions about what I do, how I do it and everything in-between.

I answered the most popular questions in this Q&A video.

💡 What Multiverse learned from 6 months of using an AI coach

A semi-case study you might want to check out.

Apprenticeship startup, Multiverse, introduced a on-demand AI coach to it’s community. So far, nearly 4,000 people have asked a total of 40,000 questions with a overall helpfulness score of 91%.

As the lines of human and digital collaboration continue to blur, we have much to learn in this space.

🔥  How to navigate overnight skills gaps

Another case study, this time from the team at Workera on navigating overnight skills gaps.

They don’t happen often but overnight skills gaps can seriously destabilise a workforce. the introduction of generative AI tools and the bag of skills and behaviours we now need for this is a prime example.

This case study covers an experiment and blueprint to deal with similar events.

How I can help you

1. My L&D toolkit stack + AI For L&D Lab with 50 zero-cost tools.
   
   

2. Work smarter with AI in L&D with my Crash Course: A 2-hour AI Crash Course for Next-Gen L&D Professionals. Join 300 + students to future-proof your skills and work smarter with practical use cases in L&D.
   
   → Access all courses at the STT Dojo.
   
   

3. Join The Cult of Thoughts 🏴‍☠️: Get access to exclusive deep dives, events and perks with our premium membership - The Cult of Thoughts 🏴‍☠️.
   
   Your investment gets you more great content and helps me keep doing this for you until the world’s tea supply runs out.
   
   

   → Exclusive subscriber-only Monthly newsletter

   → Private catalogue of toolkits and insights

   → Workshops, events and perks
   
   

4. Stalk me on LinkedIn for daily insights, tools and frameworks for modern L&D pros.

Want to reach over 3,000 L&D pros?

Learn how you can sponsor the newsletter