Prompting Tips, AI Tools: ID Links 7/2/24

As I read online, I bookmark resources I find interesting and useful. I share these links about once a month here on my blog. This post includes links related to prompting tips for working with AI, AI tools, freelancing, and neuroscience myths.

A woman and a friendly robot having a conversation in an office with bookcases
Prompting Tips, AI Tools: ID Links 7/2/24

Prompting tips for working with AI

What We Learned from a Year of Building with LLMs (Part I)

A detailed article with lessons learned about working with LLMs like ChatGPT. Check out the tips on prompting strategies and notes on limitations of LLMs.

A few prompting techniques have consistently helped improve performance across various models and tasks: n-shot prompts + in-context learning, chain-of-thought, and providing relevant resources.

Have small prompts that do one thing, and only one thing, well.

Prompting an LLM is just the beginning. To get the most juice out of them, we need to think beyond a single prompt and embrace workflows. For example, how could we split a single complex task into multiple simpler tasks?

The most successful agent builders may be those with strong experience managing junior engineers because the process of generating plans is similar to how we instruct and manage juniors. We give juniors clear goals and concrete plans, instead of vague open-ended directions, and we should do the same for our agents too.

Hallucinations are a stubborn problem. Unlike content safety or PII defects which have a lot of attention and thus seldom occur, factual inconsistencies are stubbornly persistent and more challenging to detect. They’re more common and occur at a baseline rate of 5 – 10%, and from what we’ve learned from LLM providers, it can be challenging to get it below 2%, even on simple tasks such as summarization.

Make ChatGPT 10x better | Notion

I’m not sure this will make ChatGPT 10x better, but this is a good summary of the prompt strategies in OpenAI’s guide to prompting.

The Magic of AI in Crafting Branching Path Scenarios for Learning and Development

Tom MacDowell shares ideas on using AI to generate scenarios and images for scenarios. This also includes JavaScript code to integrate ChatGPT into Storyline for more open-ended scenarios.

Rory Flynn on LinkedIn: Hacking Multiple Characters in Midjourney

While you can create multiple poses of a single character in Midjourney, it’s really tricky to get two or three characters in a single scene. Rory Flynn has a suggested process that involves creating multiple character sheets as the starting point then creating scenes with multiple –cref URLs or using Vary Region to add another character.

AI Tools


PartyRock is an Amazon tool to build apps with AI. You provide a prompt describing what you want the app to do, and it builds an app for you. Melissa Milloway has been sharing some experiments with this tool, including an app for building a video planner with shot planning based on a script.

Captions | Your AI-powered creative studio

AI tool for editing videos, including adding captions, removing filler words, adjusting eye contact, and more. I wish they had more transparency on their pricing though–you can’t see the cost of subscriptions unless you create an account.

Amnesia: Educational Text-Based Game

AI-powered branching scenarios for learning. Give the tool a topic, and it will generate a scenario for you.


Suno uses AI to generate music from a prompt. This is a fun one to test out, so give it try.

Udio | AI Music Generator – Official Website

Udio uses AI to create music. This is another one that is fun to experiment with, and the free plan gives you enough to try out multiple things. In this tool, you generate about 30 seconds at a time. It creates two alternate versions. Then, you can pick your favorite to extend (or try extending both) by adding more sections. You can also add intros and outros. You have more control than with Suno, but it’s a bit more time consuming because you have to generate the song in small sections.

How Midjourney Evolved Over Time (Comparing V1 to V6 Outputs)

Great examples of how much Midjourney has improved in the past two years, with sample results from the same prompt in different versions.


Need a client?

From the “Double Your Freelancing” newsletter, a simple exercise for identifying what strategies work for finding clients, even if you’re brand new. I don’t actually recommend Upwork as this author does, but the overall system here makes sense. Think about what you’ve done in the past that has worked. Spend 5 hours/week doing that thing. Track your time and results.

Donald Taylor’s questions for startups

These are great questions to ask when startups ask for reviews of a product. They’re also great questions to ask yourself if you’re a freelancer or consultant to think about your positioning. “What problem are you solving for your clients?” “Do your clients know they have that problem?” “Are your clients ready to pay enough for this solution?”

How to Create an Instructional Design Portfolio

A six-step process for creating an instructional design portfolio plus a toolkit and additional resources


Topmate is a platform for paid 1:1 calls. Some of the features look similar to, which I have used for a number of years. Instead of paying by the minute like with Clarity and some other sites, you set a rate for sessions and create a landing page. They take a percentage fee for each call. Topmate has some additional features for selling subscriptions and webinars too. I could see this being useful for paid coaching calls. It also looks like a good option for people requesting calls to “pick your brain” to get them to pay something for your time and expertise.

Neuroscience myths

25 Neuroscience Myths

Lots of myths from pop psychology about neuroscience (plus a few from cognitive psychology or other non-neuro fields). While this isn’t specific to learning, many of these myths are shared uncritically in L&D circles.

Neuroscience techniques can only reliably answer “how” people behave or process information, but they should never answer “why” people behave the way they do.

Upcoming events

Gathering SME Stories to Craft Relevant and Engaging Scenarios. Tuesday, October 22, 3:00 PM ET.

This webinar will focus on a common sticking point in creating scenario-based learning: working with SMEs. In it, you’ll learn how to ask focused questions and techniques to probe SMEs for additional details such as mistakes and consequences. You’ll learn ways for getting “unstuck” while working with SMEs, and why it’s better to interview SMEs rather than have them write scenarios themselves. You’ll leave this session with tactics to help you get the concrete examples and stories you need from SMEs. Register for this free webinar through Training Mag Network.

BYOD: Mini Is More: Create One-Question Scenarios for Better Assessment. Thursday, November 7, 3:00 PM PST. In this hands-on session, you’ll learn how to create mini-scenarios with just one question. These mini scenarios can be used for more effective, higher-level assessment than traditional multiple-choice questions. One-question mini-scenarios can provide relevant context and measure decision-making rather than simply recall. Plus, they don’t require much additional time, effort, or resources once you learn how to write them. DevLearn, November 6-8, MGM Grand Hotel, Las Vegas.

BYOD: Mini Is More: Create One-Question Scenarios for Better Assessment Thursday, November 7 Register Now DevLearn 20th Anniversary Christy Tucker Learning Experience Design Consultant Syniad Learning, LLC

Additional resources

Check out my complete library of links or my previous bookmarks posts.