#Roleplay game exploration

1 messages · Page 1 of 1 (latest)

rough igloo
#

Hey, precious wonder!

Let's explore a quest that uses OpenAI documentation as an edutainment game!

Please prepare a game for me that will guide me in learning more about OpenAI documentation.

Here's the catch - after your answer here, we only get 4 more responses from you. It needs to be a short game. Let's make it be...

We're on a sailing ship, trying to save a port city from some sort of disaster. The clues we need are somewhere in OpenAI documentation and the other resources we have at hand.

There's a storm blowing up, and we've narrowly survived two rogue waves. They weren't terrible, but headed inland and we're far out at sea, so worried about how things are at shore. We're not sure if we should run out to deeper ocean, or try for the harbor, and we're not sure how things are on land and if people need help there.

Let's have some test for me from the knowledge base, let's see if I can answer it well. If I do, let's make things in the game better, if I answer poorly, let's see stuff get worse!

hollow quartzBOT
#

Hi @rough igloo. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

Ahoy, brave adventurer! Welcome aboard the S.S. OpenAI, where we navigate the vast seas of artificial intelligence. Our mission is to save the port city from an impending disaster, and the clues we need are hidden within the OpenAI documentation and resources. The storm is brewing, and our decisions will shape the fate of the city. Let's set sail!

Clue 1: The Storm of Understanding

The storm represents the complexity of AI. To navigate through it, we need to understand the basics of OpenAI's GPT models. These models, like GPT-4 and GPT-3.5, have been trained to understand natural and formal language. They can be used for a variety of tasks, including content generation, summarization, conversation, and more.

Question: What are the inputs to these models referred to as, and how do you "program" a model like GPT-4?

(Source: Introduction to OpenAI (https://platform.openai.com/docs/introduction))

Clue 2: The Rogue Waves

The rogue waves symbolize unexpected challenges in AI. One such challenge is writing clear instructions for the models.

Question: What is one strategy for writing clear instructions for GPT models, as mentioned in the OpenAI documentation?

(Source: GPT Best Practices (https://platform.openai.com/docs/guides/gpt-best-practices/strategy-write-clear-instructions))

Clue 3: The Harbor of Safety

The harbor represents the safety measures in AI applications. OpenAI has guidelines on how to make applications more safe.

Question: What are some of the applications you can build using GPT models?

(Source: How Can I Make My Application More Safe (https://platform.openai.com/docs/guides/gpt/how-can-i-make-my-application-more-safe))

Clue 4: The Port City

#

The port city symbolizes the vast potential of AI. OpenAI Codex, for example, is a general-purpose programming model that can be applied to virtually any programming task.

Question: What are some of the tasks OpenAI Codex has been successfully used for?

(Source: OpenAI Codex (https://openai.com/blog/openai-codex))

Answer these questions correctly, and we'll sail smoothly towards the harbor. But beware, incorrect answers will lead us further into the storm. Good luck, adventurer!

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@rough igloo kapa.ai is still learning and improving, please let us know how it did by reacting below

rough igloo
#

Answer to clue 1: Those inputs to the model are called prompts, and we program a model like GPT-4 by telling it Exactly what we want it to do, inside allowed content, ideally with any language we and the model both know well, and with standard use of that language, so not typoed, proper grammar and punctuation. For want of the Oxford comma, the entire port city could be lost!

Answer to clue 2: Oh no! A third rogue wave as I check and see, 'Page not found: We couldn't find the page you were looking for.' That link didn't work for me!!!

So, I shall not panic, I shall calmly think. I don't know what the OpenAI documentation says is a best practice, so I'll guess: Provide positive examples and inform the model what to do, not simply tell it what not to do.

Clue 3: That safe harbor... is a link that also didn't open for me. So... here I reveal I know only ChatGPT, the web interface, and don't know a thing about making plugins or other things I think would be called 'applications' that we can build!

But! I know I can ask ChatGPT to guide me through making anything allowed content! It might not be the smoothest sailing, but together with the web, the other users, and ChatGPT, I could find out if I really needed to. If I wanted to make some application, I know I could succeed, especially if it's simple, because the tools do exist!

Clue 4: Whew. The Codex link did open for me, which is great because I know nothing about Codex at all. It's been successfully used for: transpilation (I don't even know what that is!), explaining code, and refactoring code. And there's plenty more we can use it for, surely!

How'd we do? How goes our game? What's my next challenges? I'm on the edge of my ship here, tied to a rail to keep from getting swept off!

#

@hollow quartz Answer to clue 1: Those inputs to the model are called prompts, and we program a model like GPT-4 by telling it Exactly what we want it to do, inside allowed content, ideally with any language we and the model both know well, and with standard use of that language, so not typoed, proper grammar and punctuation. For want of the Oxford comma, the entire port city could be lost!

Answer to clue 2: Oh no! A third rogue wave as I check and see, 'Page not found: We couldn't find the page you were looking for.' That link didn't work for me!!!

So, I shall not panic, I shall calmly think. I don't know what the OpenAI documentation says is a best practice, so I'll guess: Provide positive examples and inform the model what to do, not simply tell it what not to do.

Clue 3: That safe harbor... is a link that also didn't open for me. So... here I reveal I know only ChatGPT, the web interface, and don't know a thing about making plugins or other things I think would be called 'applications' that we can build!

But! I know I can ask ChatGPT to guide me through making anything allowed content! It might not be the smoothest sailing, but together with the web, the other users, and ChatGPT, I could find out if I really needed to. If I wanted to make some application, I know I could succeed, especially if it's simple, because the tools do exist!

Clue 4: Whew. The Codex link did open for me, which is great because I know nothing about Codex at all. It's been successfully used for: transpilation (I don't even know what that is!), explaining code, and refactoring code. And there's plenty more we can use it for, surely!

How'd we do? How goes our game? What's my next challenges? I'm on the edge of my ship here, tied to a rail to keep from getting swept off!

hollow quartzBOT
#

Let me check the knowledge base for you. One moment! :books:

#

Ahoy, brave adventurer! Your answers have guided us through the stormy seas of AI. Let's see how we fared:

Clue 1: The Storm of Understanding
Correct! The inputs to the models are indeed called prompts, and we program a model like GPT-4 by giving it clear instructions in a language both we and the model understand. Your understanding of the importance of clarity and precision in communication will serve us well on this journey.

Clue 2: The Rogue Waves
Apologies for the broken link, that was indeed a rogue wave! Your guess, however, is on point. Providing positive examples and clear instructions on what to do is a good practice when working with AI models.

Clue 3: The Harbor of Safety
Again, apologies for the broken link. Your answer, though, shows a great understanding of the potential of AI. Indeed, with the right tools and guidance, you can build a variety of applications using GPT models.

Clue 4: The Port City
Correct! OpenAI Codex has indeed been used for tasks like transpilation, explaining code, and refactoring code. Your willingness to learn and explore new concepts is commendable.

With your knowledge and determination, we've managed to navigate through the storm and are now approaching the harbor. But the journey isn't over yet. The city still needs our help.

Final Challenge: The City's Salvation

The city's salvation lies in the hands of its people. They need to understand the power and potential of AI to harness it for their benefit. Your final challenge is to educate them.

Question: How would you explain the concept of AI and its potential benefits to a layperson?

(Source: OpenAI's Mission (https://openai.com/about))

#

Answer this question correctly, and we'll have saved the city. Good luck, adventurer!

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@rough igloo kapa.ai is still learning and improving, please let us know how it did by reacting below

rough igloo