Unveiled only a day ago, GPT-4 is already revolutionizing the way people use AI, unleashing a myriad of mind-blowing applications that the previous ChatGPT iteration (GPT-3.5) was either struggling with before or was simply impossible to do.
While at first glance, the differences between GPT-3.5 and GPT-4 might seem subtle in casual conversations, the true distinction becomes apparent when you give it a task with high levels of complexity. The buzz generated by GPT-4 is undeniable, as people are going wild, creating an array of games and applications, and even utilizing it to perform highly advanced tasks that would typically require the costly expertise of a professional.
In this article, we’ll delve into nine astonishing ways individuals are harnessing the power of this cutting-edge AI to transform their lives and industries. And perhaps you might be inspired to conduct your own experiments.
Create Games Even Without a Coding Background
Platforms that offer no-code game creations have been around for years, but they’re still cumbersome to use and fall short of capturing users’ full ideas.
But GPT-4’s all-powerful capabilities will change this all and finally equip the masses with the ability to create games they would really want to play and see. What a perfect time to live, indeed!
1) Bring a 3D Game to Life
Javi Lopez created a 3D gaming environment using the newly-released GPT-4 large language model (LLM), which ChatGPT Plus users can now access. He disclosed right away that what he created wasn’t a full-blown game with enemies and shooting capabilities but only a virtual environment.
It is a 3D maze map with a first-person point of view, which he describes as a ‘super basic prototype.’ Think of it as a proof of concept for the language model’s massive potential to create true-blue 3D games in the near future.
This virtual environment was based on a game called “Doom,” a popular FPS game back in 1993.
The language model seamlessly generated a code for him, which Lopez ran on a code testing site. At first, GPT-4 made the mistake of creating a 2D maze instead of a 3D environment.
But by simply reminding it to create a 3D version, the LLM corrected itself and generated a revised code based on what Lopez was aiming for.
The maze was pretty amazing, which feels like it was fetched straight from the 90s gaming era. Lopez also made a promise to create a true 3D game from GPT-4, complete with weapons, enemies, and more, if the demand calls for it.
2) Create a PingPong Game in Under 60 Seconds
Pietro Schirano created a classic Ping Pong game in less than a minute using GPT-4’s more advanced abilities.
Yes, it may not be a sophisticated game, but this feat gives us a glimpse of what capabilities this LLM can provide to everyone, especially those with little to no coding backgrounds.
For this simple project, Prieto instructed GPT-4 to act as ‘DeveloperGPT,’ a type of prompting template that commands ChatGPT to act as a specific expert to help it generate more factual results.
He then asked it to create a game similar to Ping Pong and inquired about the best programming language so everyone could access his game.
GPT-4 confidently advised him to use ‘Javascript with HTML Canvas API,’ as it would allow him to run his game on a browser, making it compatible with various devices and operating systems.
The game was run on Replit, an independent development environment (IDE), which is a platform for testing and debugging codes.
3) Classic Snake Game in Minutes
Ammaar Reshi gave everyone a nostalgic feel when he used the unbelievably advanced GPT-4 to create the classic low-tech game Snake in under 20 minutes.
For those who have lived ‘a bit early’ to witness the game’s debut on the monochrome Nokia 6110 phone, this one is for you.
Reshi asked GPT-4 to generate a code for a classic Snake game, which the LLM produced complete with Javascript, HTML, and CSS.
The only hitch he experienced was that the AI chatbot struggled to generate the entire code in one go. This forced Reshi to type the word “continue” several times just to make GPT-4 complete the code for the game. A small price to pay.
He then pasted the code on Replit to test it, and as he might have expected, it didn’t disappoint. I myself was entertained by its simplicity and elegance, which is a much-needed break from the hyper-realistic games I’ve been seeing for years.
You may check out and play the GPT-4-powered Snake game here. A quick tip: After clicking the link, just press the arrow keys to start playing right away.
Create a Google Chrome Extension and Mobile App
Right now, creating Google Chrome extensions and mobile apps could be intimidating for a lot of people.
But soon, GPT-4 may open the floodgates of new creators not bound by the complexities of code creation and not reliant on marketplaces just to access these applications.
4) A Pirate-Themed Google Chrome Extension
Jake Browatzke decided to use GPT-4 in a more humorous way by creating a Google Chrome extension called ‘Pirate Summarizer’ that converts highlighted text into ‘pirate speak’.
The creator hasn’t yet provided a link to the extension, and the last time I checked, it wasn’t available on the Chrome Web Store.
Jake was quick to disclose that he has no experience in coding and has gone through several revisions before successfully creating the extension.
This concept is fun and displays GPT-4’s ability to turn even the craziest ideas into reality with just simple commands.
5) Movie Recommendation App
Now, everyone can be their own app maker, even without learning how to code!
Morten Just quickly took advantage of GPT-4’s accessibility by creating a mobile app for recommending movies. A personal movie consultant, if you may.
This app now recommends five movies for him every day, complete with trailers, and even suggests the best streaming service for each movie.
But the process of his app creation faced some bumpy roads, including bugs in the generated codes, which were eventually fixed by GPT-4. You may check the complete code here.
This experiment also showed the potential of LLMs to become an ‘ infinite mobile app store’ soon.
This means that, in the near future, you and I may not need to download anything from app stores, as we can build the apps ourselves by simply typing the features we want on an LLM like GPT-4. This is crazy exciting!
6) Create Recipes Out of Images
Soon, GPT-4 might be the next hapless victim of the notoriously hot-headed celebrity chef Gordon Ramsay, with its new-found ability to whip up recipes ー out of images.
But you may have to hold your patience as the image processing capabilities of GPT-4 is yet to come, as was clarified in its recent demo.
But an AI engineer named Sudharshan has found a way to test GPT’s visual potential through ‘Visual ChatGPT.’ It is a model that fuses the AI chatbot with visual foundational models (VFMs), allowing ChatGPT to process both text and image inputs.
In his experiment, he uploaded an image of an open fridge filled with various foods to test if the model could recognize the different items in it.
And guess what? It accurately named each one of them!
“Apples, oranges, bananas, eggs, a carton of milk, cheese, and a package of lunch meat and other items.”
He then fed these results to GPT-4 and asked it to make something from these foods.
As amazing as it is, it remains a cumbersome process. In fact, it is a far cry from what’s yet to come, which would simply allow users to upload any images of foods and wait for GPT-4 to generate yummy suggestions.
Hey, isn’t this a good business idea for a startup? You’re welcome!
7) Create a Detailed Business Plan From a $100 Budget
Struggling to make a business plan on a limited budget? A guy named Jackson Greathouse Fall tested GPT-4’s potential to solve this problem.
In his experiment, Jack asked the language model to grow his $100 budget as high as possible in the shortest amount of time ーwithout doing anything illegal.
And GPT-4’s response is hands-down impressive:
Without any prompts or instructions related to affiliate marketing, GPT-4 went straight ahead and suggested this path.
Look at the fine details of its instructions, such as purchasing domain and hosting services and setting up an affiliate website.
The crazy part of this experiment was that Jack earned revenue from the website and even attracted investors! Just take a look at the current standing of their business!
Indeed, that’s a $100 well-spent!
Founder’s note
@evan-ezquer: While I do agree with @damocles that this is an impressive feat for an AI model, the initial advice was a little generic for my taste (but that was just the starting point). Indeed, anyone starting a niche site might come to similar conclusions as GPT-4 after a short period of time. The real challenge lies in the nuances of making a website truly stand out, overcoming competition, and strategically positioning it within the market. However, Jackson actually made a lot of follow-up questions and sought ChatGPT’s advice almost every step of the way, from logo creation to topic selection, writing, ad spending, etc. And considering the $100 became $1,378, then that means that GPT-4 has the ability to give business advice in some cases. This is big!
8)Identify Security Flaws on an Ethereum Contract
Coinbase director Conor Grogan recently tested GPT-4’s capability and accuracy in detecting potential security flaws on an Ethereum contract.
Based on his experiment, the LLM instantly highlighted various vulnerabilities in the contract and even pointed out its weakest points, which hackers could exploit.
While smart contracts are already secure because of blockchain tech, AI can further strengthen them by eliminating the unseen threats hiding somewhere in their layers of code.
9) Describes the World for the Visually Impaired
Be My Eyes is a mobile app that allows volunteers to help visually impaired people in tasks that require a sense of sight through video calls.
It is a breakthrough service that opened opportunities for millions of people with visual difficulties to see the world and live normally again.
But GPT-4’s visual capabilities will catapult this service to the next level through a new feature called ‘Virtual Volunteer.’
This means that instead of a human volunteer, the LLM’s visual tech will take charge and help Be My Eyes’ clients see the world through artificial intelligence (AI).
The Virtual Volunteer will be able to talk to the users in real-time through GPT-4’s image-to-text capabilities.
In fact, OpenAI decided to first deploy GPT-4’s visual tech in this application before anywhere else, which will certainly benefit a lot of people with visual limitations.
See also:
- Supercharge Your ChatGPT Game w/ 10 Advanced Prompts
- Jailbreaking ChatGPT
- 16 Brilliant Ways to Use ChatGPT to Make Life Easier
Join our newsletter as we build a community of AI and web3 pioneers.
The next 3-5 years is when new industry titans will emerge, and we want you to be one of them.
Benefits include:
- Receive updates on the most significant trends
- Receive crucial insights that will help you stay ahead in the tech world
- The chance to be part of our OG community, which will have exclusive membership perks
