AI-Powered Interactive Kiosk (GPT-3)

Show & Tell
1

AI Kiosk

I’d like to share a new demo inspired by the Museum of Flight in Seattle, USA and by a recently released OpenAI-GPT IA that makes it possible for Intuiface to engage the powerful GPT-3 AI model by OpenAI.

The demo consists of three main components: a user interface (UI), an Excel file that combines user preferences into a meaningful GPT-3 prompt, and a custom IA that communicates with the OpenAI API to retrieve the answer.

The UI allows users to select a topic and language, and to choose the complexity level of the answer using a slider that ranges from 0 to 10. The slider also features a converter that adjusts the length of the answer based on its complexity level.

Once the user’s preferences are collected, they are stored in an Excel file, which forms the basis of the GPT-3 prompt. To ensure that the AI model produces answers that match our requirements, we pass certain parameters to it in a structured format. Here is what the full prompt looks like:

Please explain a topic following a specific prompt format that uses brackets to separate various parameters. Here is the format:
[Topic], [Language], [0-10], [Max Words]
Let me explain each parameter:
[Topic] is the subject that I need you to explain.
[Language] is the language that the explanation should be translated to.
[0-10] is the desired level of complexity of the explanation. The complexity level is indicated on a scale of 0 to 10, where 0 corresponds to a very simple explanation suitable for elementary school students, 5 corresponds to an explanation suitable for middle-school students, and 10 corresponds to a college-level explanation.
[Max Words] is the maximum number of words allowed for your explanation. Please don’t exceed it.

The OpenAI-GPT IA is a custom API developed by @seb for the Intuiface community. It enables users to access specific AI models from OpenAI, submit custom prompts, and set a limit on the number of tokens (or words) in the response.

davinci

By limiting the amount of data transferred in each query, users can control the cost of the service while still benefiting from the power of OpenAI’s AI models.

Feel free to download and test this XP.

Please note that to use this XP and customize it, you will need to sign up for the OpenAI API platform and obtain your own secret API key. Once you have your API key, simply switch to Edit mode in Composer, locate the Excel database file called XLS_PromptMaker under Interface Assets, and paste your API key into the appropriate field. Enjoy!

4 Likes
2

Keep it up it as always! :heartbeat: I shall wait for you to create an example with GPT-4 with images taken on a webcam from an experience.

1 Like
3

@tosolini, just an awesome expansion of what we originally posted to the Marketplace. You constrain the topic so the user cannot ask anything off-topic, but you also give that user lots of control to 1) create consumable content, and 2) have some fun.

One note: @tosolini’s experience uses the .NET version of the GPT interface asset. (It’s the IA named “OpenAIGPT”). This means you must use the Windows version of Play Mode. It will not work with Player Next Gen.

2 Likes
4

Thanks @Promultis and @geoff. As the OpenAI models evolve (and new ones may be released by other competitors), it’ll be fun to integrate them with Intuiface to satisfy emerging use cases. In this specific instance, I found myself spending quite a bit of time fine tuning the prompt to make it work correctly. So Prompt Engineering is very much part of the XP design with AI.

2 Likes
5

Was hoping to play with this XP but am having issues getting it to work. I’ve entered my API key and the XP starts and lets me select the preferences but once the request is made composer/player (have tried both) exits after about 1-2 seconds and I get the message “Play mode closed due to an unexpected error. You can try again after clicking the OK button.”

I will note that I get the same result when trying to use the “Playing with Generative AI - Windows” XP from the Intuiface Design Team.

Any ideas what my issue is?

thanks
Brad

6

This might be more of a @Seb question, but OpenAI made changes to their API for new models, and the IA of these demos might be outdated. I experienced the same crashes when I tried to use GPT4 using an IA that could only support older models (like text-davinci-003).

1 Like
7

Hi Brad @emphelp ,

I just tested the Playing with Generative AI - Windows sample with my API key and didn’t get any issues there.
I believe we do have a known crash if you try to use an API key that doesn’t have any token available, could that be your case?
Since we are going to replace this GPT 3.5 IA with one that supports GPT4 soon, and that crash was fixed, we are not planning to fix the issue in the 3.5 version.

I’ll share with you an example for Windows that works with GPT4, you can thank Paolo for supporting the development of that version :wink:
This is not an official Intuiface component, so it is delivered as-is.

If you have any question or issues with your XP, please contact our support team.