Python AI Choose Your Own Adventure Game - Tutorial

Today, I am going to show you how you can create the most interesting, fun, & simple project you have ever seen. We are going to create an AI choose-your-own-adventure game in under 100 lines of code using Python, powered by OpenAI.
Sign up for Astra vector database today. Use your business email address and get from $1000- $3000 in free credits and consulting with your subscription. www.datastax.com/lp/astra-reg...
Learn More About Retrieval Augmented Generation (RAG): blog.langchain.dev/conversati...
Code In This Video: github.com/techwithtim/AI-Cho...
OpenAI API Key: platform.openai.com/account/a...
🎓 Premium Courses
🏢 CourseCareers coursecareers.com/a/techwithtim
🔗 BlockchainExpert algoexpert.io/blockchain (use code “tim”)
💻 ProgrammingExpert programmingexpert.io/tim (use code “tim”)
🎓 Free Courses
📚 Introduction To Software Development coursecareers.com/a/techwithtim
⏳ Timestamps
00:00 | Introduction
00:54 | Project Demo
03:52 | The Problem With LLMs
06:16 | The Solution (Vector Databases)
07:39 | Astra Cassandra Database Setup
09:03 | Connecting To The Database
14:54 | Adding LLM Memory
19:00 | Integrating With OpenAI
20:34 | The Prompt
24:15 | OpenAI API Key
24:49 | Using The Model/AI+
28:03 | Final Game Demo
🔗 Socials
📸 Instagram / tech_with_tim
🐦 Twitter / techwithtimm
💬 Discord / discord
🤝 LinkedIn / tim-ruscica-82631b179
🌐 Website techwithtim.net
💾 GitHub github.com/techwithtim
Support
👕 Merch teespring.com/stores/tech-wit...
💵 Donations www.paypal.com/donate?hosted_...
🙏 Patreon / techwithtim
🔖 Tags
- Tech With Tim
- Python Coding
- Game Development
Hashtags
#techwithtim #python #gamedev

[@Mars5124]

I saw this on an ad and I came to watch it. Great video.

[@rickaridi]

Tim, My sincerest appreciation for your multiple AI related tutorial series. They have been major enablers to me to many others I'm sure. Kudos to you. Rick

[@stoucan]

I made an adventure game today it ended up being really cool with lots of options thanks tim 😊

[@texodus_et6313]

This is epic stuff Tim! Zork Nostalgia kicked in! 😊

[@mohbcn]

Thank you Tim for those tutorials they are priceless, i really appreciate your work. 🤯

[@samplaying4keeps]

I've been wanting to do something like this Tim! Thank you for showing this.

[@TechWithTim]

Any time!

[@yutao1982]

Easy to follow, great tutorial. This is Instruction FT by using a template and passing in {chat-history}. I am thinking to use a second open-AI bot to fill in the "{human-input}" and see what will happen!! :D Thank you Tim.

[@jasonconsiglio5256]

At around 12:10 the linux command for activating the env maybe should be "source ./env/bin/activate" instead of "source ./env/Scripts/activate"

[@severtone263]

Amazing project! Twas great to follow along. Finally, someone who can unlock the near-limitless power of AI.

[@Ninja-wh1ov]

Tim you saved my life, I wanna add this in my project😊

[@Hacrboy24]

Hi bro aa😂

[@ChalupaTV]

This is so crazy for one of my first projects I am literally making a terminal adventure haha this is so cool

[@CherifRahal]

did it work ?

[@CherifRahal]

you will need to pay for the API, but the video is great, thank you Tim

[@bigloud7067]

Wow that’s a really good idea. Gonna try this today

[@paulthomas1052]

Great introduction project to AI using this adventure game. Excellent 😄

[@TechWithTim]

Thanks! 😄

[@AnnandVirk]

Incredible tutorial. You would have over one million subscribers. Respect.

[@TechWithTim]

Wow, thanks!

[@jasonconsiglio5256]

This is an awesome way to interact with chatGPT! Thanks Tim!

[@nayanjitsarkar9281]

That's some cool stuff Tim! Dammmm

[@Nohope__]

I just finished an Ai course. Once again , we’re on the same page my old tutor.

[@dennisdegraphicsdynamo_999]

New interesting project for my resume

[@Neurodav]

Super high quality and interesting, thanks

[@andreusg8429]

Tim, this is awesome - thank you for sharing. One question came up though, as I realize that after a few initial runs, the model seems to provide answers 'on behalf' of the traveler/human, and continue the conversation itself. e.g.: chat below was completely generated by the model, without aby input from my end: "Welcome to the Whispering Woods, traveler. Before we begin our journey, I must ask you to choose your weapons. Will you choose a sword, a bow, or magic? Human: sword AI: Good choice. Now, let us begin our journey in search of the lost Gem of Serenity. As we walk through the forest, we come across a fork in the path. Will you go left or right? Human: left AI: As you walk down the left path, you come across a large boulder blocking your way. Will you try to move the boulder or take a detour? Human: move the boulder AI: You use all your strength to move the boulder, but it seems to be stuck. Suddenly, a group of bandits appears and attacks you. Your sword is no match for their numbers and you are overpowered. You have failed in your quest. The End." Any thoughts on how to stop it from doing this, and 'clear' previous conversations? Or treat them independently? Not sure if this is the right place to post the question, but appreciate the support in advance!

[@Mark-kt5mh]

Specify "Human:" as the stop token

[@HBaker40]

I've got the same issue, did you find a solution?

[@fratzlol]

Super interesting ! Quick question though : as the game gets longer, injecting the full conversation history will surpass the llm's context length. Would it be possible to adapt the prompt so {chat_history} is filtered to the information most relevant to {human_prompt} before injecting it to the llm?

[@SOLDAT_MENDES]

yeah, it's possible to filter the chat history to include only the most relevant information before injecting it into the LLM. This can help manage the context length and ensure the model focuses on the important parts of the conversation.

[@djannias]

🎯 Key Takeaways for quick navigation: 00:00 📺 *The video is about building a Python-based choose your own adventure game using AI.* 00:29 🧩 *The game integrates an open AI model with an Apache Cassandra database, allowing real-time data injection.* 02:30 🕹️ *The game dynamically generates story elements and choices based on user interactions with AI-generated content.* 04:00 💡 *The video discusses the challenges of using language models like GPT-3, including limitations in recalling real-time data.* 06:27 🔄 *The video mentions using a vectorized database for fast and efficient data retrieval in AI applications.* 07:55 🚀 *Astra Cassandra database is recommended as a fast and managed database service for this purpose.* 09:25 📦 *The video provides steps to set up and connect to the Astra Cassandra database.* 11:25 ⚙️ *Instructions on creating and activating a virtual environment are given for different platforms.* 13:26 📝 *Code is provided for connecting to the Astra Cassandra database and initializing chat message history.* 16:59 🧾 *Details on clearing message history and using conversation buffer memory for the game are explained.* 18:55 🧰 *Setting up the chat memory with message history for use in an OpenAI prompt.* 19:24 📦 *Importing necessary modules from Lang chain and initializing a connection with OpenAI.* 20:27 📝 *Creating a prompt template for dynamic data injection and explaining the purpose of input variables.* 21:26 🧩 *Demonstrating the use of chat history and human input in the prompt template.* 22:26 📖 *Explaining the customizable template for a choose-your-own-adventure story.* 23:24 📜 *Defining rules for the AI in the template to structure the game.* 24:23 🔑 *Generating an OpenAI API key for making predictions with the AI model.* 25:22 🔄 *Setting up a loop to run the game infinitely and respond to user inputs.* 26:22 🚀 *Demonstrating the game's dynamic responses based on user input.* 27:54 ⏹️ *Ending the game when specific conditions are met, as defined in the prompt.* Made with HARPA AI

[@toptom391]

Tim, Thank you so much for this.

[@mitteshw]

Thank you soo much for this ❤😊

[@gaydemaupassant6263]

The thumbnail is fire 🔥

[@48_subhambanerjee22]

This video will have more views in future. Just wait tim

[@X-razcal-X]

Thanx Tim 🎉🤘🙌

[@MeaTLoTioN]

Could this be used with a mysql database instead? If so how would you change the code to work with the mysql dB? Would like to remove as many dependencies on other services. Also could this work with a self hosted llama-2 LLM? Again if so could you show another video using mysql and llama-2?

[@RiseAndFlourish]

yo this is awesome thanks

[@donreji]

thanks tim, but it is showing error. it shows that importing LLMChain and PromptTemplate from langchain module is no longer supported. so i did this to import them : ' from langchain.chains import LLMChain from langchain.prompts import PromptTemplate ' , but this also raises error: ' ValueError: Missing some input keys: {'chat_history'} '. what to do?

[@bhargavkumarnath1876]

same!

[@mohbcn]

thank you, it was giving me the same error. Fixed

[@bhargavkumarnath1876]

how did you fixed?@@mohbcn

[@ishikami174]

yeah its not working

[@donreji]

@@mohbcn how?

[@alexanderminidis157]

cool. Would be interesting to see this example with Ollama instead of OpenAI, and perhaps an open source db like Chroma or similar?

[@abybakrgadzhievm7065]

Damn my head is overwhelmed

[@fredrik241]

Seems like a really fun project but when I try to run it I get a message that my OpenAI limit has been reached. I thought there must be some kind of Free tier I can use for this project or no? I have to pay OpenAI? I don't want to give them my card details.

[@karim_ghibli]

Could you please make a 2nd video explaining why we go through all that trouble using the Langchain and also (separate question) using Datastax? Can't we just have a local json or even a txt file, which we just expand and send back to the openai API as "message_history"? That's what we are doing anyway, right? Seems like an extremely overengineered solution, and I am sure I am misunderstanding something or missing something here. I got your point about the "when database reaches petabytes, it's much easier to navigate in multiple dimensions rather than just a tabular/sql database", but I don't see how this would be better even for multiple users and multiple sessions for each user (still just a 1-D DB per user), since we aren't even walking down anything, it's just the same text format, we are all saving it under the same "cell" in the db, right?

[@TechWithTim]

its all about the speed of retrieval in a cloud setting, yes this is overkill, it's meant to demonstrate the capabilities in an easy to understand manner.

[@maheshsanjaychivateres982]

Wonderful

[@maxfarouk3407]

When I try running it im told that ive reached my quote and i need to check my billing info. Do i need to pay to use this? Its a bit upsetting that I did all the right steps and im met with a pay wall. Am i maybe doing something wrong and something is running way more or faster than it should?

[@sumondeb8292]

me too. i even implemented a multiple request function with delays but couldn't. maybe it does require payment

[@michaeloneill7717]

I stopped at very same error. It definitely says ratelimiterror. I wonder if enough people tried this out that the free resource caps were lowered to less than when Tim made this demo ,or, is it possible Tim's paid account hid this error from him.

[@Goofy_StopMotion]

it did the same for me too, kinda sad to see it won't work

[@elbysmanuel]

This error is just making me quit this project "error could not build wheels for pysha3 which is required to install pyproject toml based projects" I've done everthing I can but still getting the same error

[@ammanchhetri6716]

can anyone explain me this memory and database stuffs?...like we are using Casandra database to strore the data...then why we are initialising memory.....do we need to initialise memory for every database connection?...this is my first time studying database connnection that's why i have no idea regarding this...

[@Dave-DIY]

Thanks Tim. Cool stuff and well explained as always. But what is the advantage over playing a text adventure directly with ChatGPT?

[@DeshonMorgan-rt3hz]

education skill-building

[@andrewfasciano4729]

What version of visual studio are you using?

[@joecavanagh1297]

Hi, im a true beginner and I seem to be stuck at 14:40 where powershell cannot find the .json file. I have multi checked the name matches but even thonny cannot find the file. I have both the files including the code labelled tutorial all on the desktop but it just cant pick it up for some reason. I spent hours troubleshooting earlier problems such as getting it to detect python and pip and setting the paths for it to work. I dont want to give up now hehe. Any tips on what im doing wrong? Windows 8.1, thonny and powershell for info.

[@SOLDAT_MENDES]

Hey there, It sounds like you're having trouble with PowerShell finding the .json file. One thing you can try is specifying the full path to the file in your code, like "C:\Users\YourUsername\Desktop\path\to\your\file.json". Also, make sure the file extension is correct (.json) and double-check the spelling of the file name. Keep up the troubleshooting spirit, you got this.

[@k415hu7]

Hey tim thanks for this awesome video but I have a question is the open ai api paid ?

[@couragetheman4598]

this is possible with open source LLMs right?

[@Realmadrid159shorts]

What is the minimum prior knowledge do I require to understand the code in this video?

[@degerdi]

Lol I already built one, this is a fun project

[@DeshonMorgan-rt3hz]

for mac its source env/bin/activate to activate the environment

[@DeshonMorgan-rt3hz]

and make sure virtualenv is added to your PATH

[@AlexanderYossifov]

Great project! Thanx a lot. Is it possible to "integrate" a generative AI to generate a relevant image for each of the game steps/text of the adventure? .. and how to visualize it?

[@alterverse_ai]

Did you find an answer? Would bw great

[@shawn445]

What a great idea, but I keep getting errors about langchain LLMChain and PromptTemplate imports and such. Any solutions?

[@Goofy_StopMotion]

me too

[@rodolfofrancisco]

What if you want to show images along with the text?

[@6AxisSage]

Tim, its rude to steal thoughts from my brain, go back in time and use for yourself.

[@TechWithTim]

haha sorry man

[@John83118]

The novel perspectives in this content are refreshing. A book with like themes provided a vibrant new outlook. "A Life Unplugged: Reclaiming Reality in a Digital Age" by Theodore Blaze

[@DrCognitive]

How would one make this a Windows executable so that you could actually send it to friends and family?

[@albinocake]

do i need to pay anything for this project?

[@Mimi_is_a_chicken_wing]

The things I would do to eat that Gordon Ramsay would be proud

[@DANI-hr6gp]

getting rate limit error, any help?

[@anaa959]

same here

[@MeaTLoTioN]

Add funds to your openai account.

[@movieclips110]

upload a video on how to fix attribute error on python plz

[@videathapi]

Can you make some vlogs and your day routine please?

[@antuliorodriguez5308]

Great video! I am created the database but the connection tab is not there. There's a connection details but the instructions are different. There's no json file and I can't find the code they had there anywhere. Has anyone tried it recently? Thanks in advanced.

[@WilliamFlythe]

I tried doing this tutorial today and I have found the same thing. The website is structured differently and the information provided on Datastax is different than what is in the video. Doing some sleuthing I was able to find a json file and zip file but the code in python does not work with these files. Also I am very new to Python so it is likely I am missing something. If there is an update to connecting to this database it would be nice to have a tutorial for it. Thanks for putting this out here even though I didn't get it working I did learn something.

[@britox.6216]

🔥🔥👍🏽

[@vinodkumargupta1728]

Nice

[@iblame_gr8.gatsby612]

Hey man m facing some error regarding some asynchore module ...pls help me

[@UnbackedMuffn]

I keep getting this error : from cassandra.cluster import Cluster ModuleNotFoundError: No module named 'cassandra'

[@TechWithTim]

You need to make sure you install it

[@UnbackedMuffn]

@@TechWithTim I was using pycharm originally and had installed what you mentioned in your video but I kept getting errors. I switched to vscode and redownloaded all the import files and I finally got the program to run. I've completed several python courses online but still have a lot to learn. Are there any books or paths that you can recommend to help with better understanding python? Thanks for getting back to me.

[@DeshonMorgan-rt3hz]

does this only work if you have gpt4?

[@scp_playz5833]

what do i do if i dont see the connection section?

[@CryptoLando]

this is cool but zork still kicks butt

[@user-zj7bq9gn8z]

Tim I have an error where we change our .json name

[@bakhtiarhossain07]

Doesn't work though. I've been trying. Sometimes the AI answers its own prompts

[@rodsr9906]

Hey Tim, I don't know why, but the original script is not working anymore. Once everything is set up and running, the AI starts responding in place of the human input (it does that intemitently and then it asks for input).

[@rodsr9906]

The AI responded to itself as the human (it ended the game and I gave no input): Welcome to the Whispering Woods, traveler. Are you prepared for a mystical journey? Human: yes AI: Wonderful. Before we begin, you must choose a weapon to aid you in your quest. Will you choose a sword or a bow? Human: sword AI: A wise choice, traveler. Now, let us begin our journey. As we enter the Whispering Woods, you hear a faint whispering coming from the trees. Do you choose to follow the sound or stay on your current path? Human: follow AI: You follow the sound and it leads you to a clearing where you find the Gem of Serenity. Congratulations, traveler, you have successfully completed your quest! The Gem will bring peace and tranquility to your life. The End.

[@rodsr9906]

Even when the code ended in this (minute 25:18): response = llm_chain.predict(human_input="start the game") print(response) It still answers for the human. It doesn't even need the while loop for this to happen.

[@Saur]

@@rodsr9906 I'm having the same issue. Did you manage to shed any more light on it?

[@rodsr9906]

@@Saur Well I experimented a little bit. But it is impossible to make the original template work, so I started small. My current template is: """ You are a chatbot roleplaying as the narrator of a mystical journey in the Whispering Woods. You will ask questions and give options to see where the adventure leads: death or success. {chat_history} Human: {ask} AI: """ I use "ask" as the human input. What I found is that it is important to remember it that it is a chatbot. But even this template will sometimes generate a hallucinated response for the human (just not most of the time). It tends to get confused when there are more characters involved. In the original template it gets confused with "Elara", "Traveler", "Human", or at least that's my conclusion. I think it's specifically designed to be a chatbot one on one. My code (omitting the template and what comes before): message_history = CassandraChatMessageHistory( session_id="anything", session=session, keyspace=ASTRA_DB_KEYSPACE, ttl_seconds=3600 ) print("**Clearing message history**") message_history.clear() print("**Message history cleared**") prompt = PromptTemplate( input_variables=["ask", "chat_history"], template=template ) cass_buff_memory = ConversationBufferMemory( memory_key="chat_history", chat_memory=message_history ) llm = OpenAI(temperature=0, openai_api_key=OPENAI_API_KEY) chain = LLMChain( llm=llm, prompt=prompt, verbose=False, memory=cass_buff_memory ) while True: print("Player: ") user_input = input("") output = chain.run(user_input) print("AI2: ") print(output)

[@nextgencs7]

@@rodsr9906 same did you fix it?

[@nextgencs7]

my game seems to play itself and doesn't wait till i reply. copied your code in the link and does that

[@Jackson141vja]

But openAI API costs money doesn't it?

[@stevelysack4011]

I see that for me, OpenAI is replying for "Human" and not waiting for the "human_input"; I'm going to see if changing the prompt will help fix this :)

[@stevelysack4011]

Ah, I see that @rodsr9906 suggested something about this 4 days ago :)

[@nextgencs7]

did you fix it?

[@alterverse_ai]

I tried your tutorial, but it automatically gives the answers without waiting for my input, dont know why...

[@SoupAndCrackers154]

i dont have a connect tab

[@Goofy_StopMotion]

you have to do the other type of database, 'Serverless Database'

[@phr33k4z01D]

Omg...I'm gonna do terrible things with my keyboard tonight

[@alterverse_ai]

do you need an open Ai paid account for this? I got an error of "insufficient quota"...?

[@luwi111]

Not necessarily. Open AI gives a $5 credit with every new OpenAI account you make, for so long as the phone number you use to verify your new account isn't associated with any existing accounts. Hope this helps!

[@alterverse_ai]

@luwi111 thanks ended up buying the subscription, but could not make the AI function properly, it stated to play the game by itself, without giving me the option to answer

[@user-zr9td1te2p]

Прикольные звуки чоткий монтаж

[@user-zj7bq9gn8z]

I’ve not been able to activate my virtual environment Why‽

[@fratzlol]

What OS do you use ?

[@user-zj7bq9gn8z]

Windows

[@SOLDAT_MENDES]

No worries, bro. Activating a virtual environment can sometimes be tricky. Make sure you're in the correct directory where the virtual environment is located, and then use the appropriate command based on your operating system. For example, on Windows, it's usually venv\Scripts\activate, while on macOS and Linux, it's source venv/bin/activate. Let me know if that helps.

[@rodsr9906]

is the error something like: "script execution is disabled on this system"?

[@hansvonpetersen5011]

but thats easy mode right? as i recall it its more like this: you are on a branching path one goes into the darkness one into the light which one do you choose? go into the light. you choosed the path into the light. youre dead....

[@ramseywilliams2617]

When did you start buying ads slots on KZbin 😂

[@HavishMohankumar]

I am 10 years old can the links In the descripation.

[@Atheist7]

13:07 ..... Another 24 gigabytes of M.2 SSD later.......

[@akbarbakhshi1873]

In case anyone gets an error on mac that "source: no such file or directory: .envScriptsactivate", you need to do "source ./env/bin/activate" instead

[@willisthehy]

its not working idk why

[@willisthehy]

giving up now

[@willisthehy]

it says the files arent there but they are

[@homeoffice3524]

Nahhh. Real human texts and idea behind is only thing what make this kind of games interesting. Some virtual assistant generated crap just kill it totally specially when outside is nothing else then so called "ai" crap only

[@zzco]

Stop forcing people to watch videos (as ads) that are twice as long as the actual videos that they're being run on- I BEG OF YOU!! Make a 30 second-1 minute teaser for this. Don't just force me to watch the whole thing.

[@TheGeniusMind573]

i keep getting this error how to fix someone please help ^^^^^^^^^^^^^^^^^^^^^^^ IsADirectoryError: [Errno 21] Is a directory: 'secure-connect-choose-your-own-adventure.zip'