Barrier Analysis plugin for
ChatGPT

Barrier Analysis plugin for ChatGPT

I've been building ChatGPT plugins to support a variety of reasoning processes. In this article, I'll run through a plugin I've been developing for behavioural barrier analysis. It's still a prototype, but it's already quite handy.

No alt text provided for this image

But before getting to the demo, first some context.

What is behavioural barrier analysis?

Many solutions require people to adopt or change a behaviour. Whether it's encouraging people to recycle, engage in a learning experience, adopt a healthier lifestyle, or use a new software tool, understanding the 'why not' can be as important as the 'how to'.

At its core, behavioural barrier analysis is a systematic approach to understanding the obstacles that prevent individuals or groups from adopting a desired behaviour. These barriers can be psychological, social, economic, or environmental, just to mention a few. By identifying these barriers correctly, we can develop strategies to overcome them, making it easier for people to change their behaviour.

Easier said than done.

Why build a ChatGPT plugin?

Barrier analysis is an important step in a solution design process. If it's not done properly (or at all), we risk wasting a lot of time and money trying to solve the wrong problems. And we risk making matters worse.

But barrier analysis is difficult. For any desired behaviour, the number of possible barriers to consider can be overwhelming. Not only that, but the intricacies and interconnections between these barriers can be complex and multifaceted. And not only that, but relevant barriers are often different for different people.

It's a lot to consider. Even with simplifying models and frameworks, it's still a lot to consider. When faced with a large number of possibilities or a high degree of complexity, people tend to become tired, bored or overwhelmed. Because of this, barrier analysis is often limited or biased.

As you'll see below, a ChatGPT plugin makes it possible to use GPT-4 (currently the most capable LLM) to lend structured reasoning and imagination support. The user doesn't need to know how to prompt and they also don't have to have a deep pre-existing knowledge of different frameworks. This helps to make barrier analysis more accessible.

Also, a plugin works out much, much cheaper than an app using the API and it's possible to chain multiple plugins without having to worry about integration.

The demo

Have a look and let me know what you think in the comments.

No alt text provided for this image
No alt text provided for this image
No alt text provided for this image
No alt text provided for this image


No alt text provided for this image


No alt text provided for this image
No alt text provided for this image
No alt text provided for this image
No alt text provided for this image


Olga Feingold

Leadership & Org Dev Leader | Director @ NRG Vivint | MBA Fellow | Driving Team Performance, Strategic Growth, and Organizational Excellence | Partnering with Executives to Solve Complex Business Challenges

1 å¹´

Incredible!! Thank you for sharing!

Daniel Taylor

Growth Instigator | Learning & Leadership Development Specialist | Neuroscience-based Coach

1 å¹´

Very cool! This is helpful and seems fairly easy to use. I wonder about how much the answers change based on the model. I also wonder about in the final step being able to transition into brainstorming about ways to engage the barriers. And finally I wonder about the ability to tweak some of the outputs, ie utilize the conversational aspect of the LLM to hone in on particular aspects of a model and ask for different responses or for it to include an additional piece of information. This is really promising though!

Sparky Witte

Drive more healthy actions with behavioral design ? Training and consulting for healthcare marketing and CX

1 å¹´

This is fascinating, Justin Germishuys. Thanks for sharing! Can you say more about what instructions you have included in the plugin or what process you are asking Chat GPT to follow? Or maybe you already have an article where you have documented your process that you could share. Robin Krieglstein - I know you would be interested in this.

Justin Germishuys

Founder at Germinotion | Combining AI with learning and behavioural design | Learning & Behavioural Designer

1 å¹´

Robert Meza, since you post quite a lot about reasoning support frameworks, I thought you might find this interesting. I'd be interested to hear what you think. Sparky Witte, I said a few months ago that I'd tell you more about this AI barrier analysis project. Sorry I forgot, but here you go.

要查看或添加评论,请登录

Justin Germishuys的更多文章

  • Evaluating with AI: The Critical Role of Anchoring

    Evaluating with AI: The Critical Role of Anchoring

    If I ask you, "How crunchy is popcorn on a scale from 1 to 5?", how would you answer? Take a moment to answer this. Do…

    38 条评论
  • Role Prompting: Hero or Hype?

    Role Prompting: Hero or Hype?

    There has been no shortage of posts, cheatsheets, and carousels on social media on the topic of prompting. And, mostly,…

    4 条评论
  • ChatGPT and the Curious Case of Alexei Romanov

    ChatGPT and the Curious Case of Alexei Romanov

    My wife and I thoroughly enjoy watching whodunits. It's not uncommon for one of us to pause a mystery show at the end…

    6 条评论
  • Experiments in Conceptual Analysis with ChatGPT: Machine Assisted Cognition

    Experiments in Conceptual Analysis with ChatGPT: Machine Assisted Cognition

    Conceptual analysis is an undervalued skill. Working in learning and behavioural design, I've seen projects spiral into…

    4 条评论
  • Thought Experiment about ChatGPT with ChatGPT

    Thought Experiment about ChatGPT with ChatGPT

    Context for thought experiment: A mysterious bright light engulfs the earth. After it dissipates, people discover that…

    3 条评论

社区洞察

其他会员也浏览了