Minecraft’s Quirky Gameplay and Open World Building Possibilities Remain Unmatched

Minecraft remains remarkably popular a decade or so after it was first released, thanks to a unique mix of quirky gameplay and open world building possibilities.

A New Twist: Oasis, the AI-Powered Minecraft Knock-Off

A knock-off called Oasis, released last month, captures much of the original game’s flavor with a remarkable and weird twist. The entire game is generated not by a game engine and hand-coded rules, but by an AI model that dreams up each frame.

Oasis: A Game Built on Transformer AI

Oasis was built by an Israeli AI startup called Decart in collaboration with Etched, a company that designs custom silicon, to demonstrate the potential of hardware optimized to power transformer-based AI algorithms.

Oasis uses a transformer AI model, similar to the one that powers a large language model—only trained, apparently, on endless examples of people playing Minecraft, to dream up each new video frame in response to the previous one and to user input like clicks or mouse moves. Oasis is similar to a video-generating model like Sora except a user can control its output.

Exploring Oasis

You can play Oasis online for free, and it is both fascinating and surreal to explore. Besides harboring bizarre artifacts, like misshapen livestock and stairs that go nowhere, the game has an amazing, Inception-like quality. Because each frame is generated based on what the AI model imagines should come after the frame it currently sees, the in-game world is never entirely stable, and will gladly shift and morph with a little nudging. If you stare too closely at a texture, for example, when you look up again, the block world in front of you may be completely different from the one you last saw.

Customization and Viral Fame

It’s also possible to upload your own image for Oasis to work with. I tried adding a photo of my cat, Leona, and the game turned her into a beautiful blockish landscape (sadly not a feline character in the game, but hey …).

Oasis has become a viral hit with people exploring ways to get its AI engine to hallucinate new environments. Sometimes, it can even be tricked into teleporting you to a dark moonscape resembling The End of Minecraft. It’s telling that this generative AI project is not entirely original, but rather seems to be a bizarro knock-off of the world’s most popular game (it was trained on an open source Minecraft dataset from OpenAI).

Industry Insights

“People are trying to teleport into different worlds and speed run,” says Robert Wachen, chief operating officer at Etched. “It’s one of the main reasons it went viral.”

The AI approach taken with Oasis is too inconsistent and uncontrollable to be useful for a conventional game, says Julian Togelius, a computer science professor at New York University. Generative AI has future potential for controlling in-game characters and perhaps generating scenes or worlds, he says, but it is still early. “It is very interesting and impressive technology, but at the moment it is an answer in search of a question,” Togelius says.

Frank Lantz, a game designer and director of the department for game design at New York University, says that Oasis seems to be caught in a kind of uncanny valley that prevents it from being genuinely fun to play. But he suggests that an enterprising young game designer might well figure out a way to turn that game into one people love. “This is so obviously cool and interesting,” he says.

Conclusion

Oasis is an intriguing experiment in generative AI and its potential applications in gaming. While it may not be a fully functional game, it has captured the imagination of many and sparked interesting discussions about the future of AI-powered gaming.

FAQs

Q: What is Oasis?

Oasis is a Minecraft-like game built using a transformer AI model that generates each frame of the game based on user input and the previous frame.

Q: How do I play Oasis?

You can play Oasis online for free by visiting the official website and interacting with the game using your mouse and keyboard.

Q: Can I customize Oasis?

Yes, you can upload your own images to Oasis and see how the AI model interprets them and incorporates them into the game.

Q: Is Oasis a game that can be played seriously?

Oasis is more of a curiosity than a fully functional game. While it can be fun to explore and experiment with, it lacks the depth and polish of a traditional game.

There are two ways to make money. You can exchange your time and expertise for a salary. Alternatively, you can buy an asset or create a system that will make you money even while you’re away.

Having a job that pays you a fixed salary is great. But creating passive income is where you want to invest more energy. In this guide, we’ll show you how to leverage automation to make passive income conveniently. Let’s dive right in.

Blogging – Automated Content and Monetization

Blogging is an excellent way to automate your passive income. All you need is a website created through a free platform like WordPress and content that will attract people to your blog.

You can create the content, outsource the work to freelancers, or use an AI content creation tool. Regardless, the goal is to publish high-quality, informative content.

Use keyword research tools to optimize your content for high-volume keywords. Crucially, focus on evergreen content that can make you money in the long term. Then monetize your blog through paid ads or affiliate links.

If you have great content, your blog will require little maintenance. Yet, you will get paid every month. Neither ads nor affiliate links require a lot of work. Once you automate the process, all you need is to wait for profits to kick in.

Automated Passive Income Apps

Passive income apps come in all shapes and sizes. The best apps are automated so that you don’t really do anything to make money. Take the app Honeygain as an example. It pays you to share your unused Internet bandwidth.

Honeygain works with companies that require loads of Internet bandwidth TO find leads and monitor their marketing efforts every month. The app keeps your data safe. It’s also legit.

Honeygain pays you $10 to $50 per month for your unused Internet bandwidth. It’s not a lot. But you can make this automated side hustle worthwhile by installing other apps.

For example, you can install MobileXpression, Savvy Connect, or Nielsen’s Mobile Panel app. These apps stay on your phone to monitor your Internet traffic – favorites sites, social networks, and games. And they pay you up to $100 per year.

Staking Crypto

Staking involves committing some of your cryptocurrencies to help keep a blockchain such as Ethereum or Solana secure. When you stake coins, the blockchain uses them to verify transactions.

Verifying crypto transactions is important for keeping the system safe. That being said, blockchains also mint new coins every time a block of transactions has been added to the system.

These new coins are awarded to crypto owners. Now, not every blockchain supports crypto staking. For example, Bitcoin uses Proof of Work to mint new cryptocurrencies.

Blockchains that support staking use proof of stake. The best coins for staking include Ethereum, Cardano, Solana, Polkadot, and USDT. These coins pay you 3% to 20% per year as a reward.

Drop Shipping

When you think about it, dropshipping is centered on automation. Customers place an order on your website. The system sends a message to your fulfillment partners. They ship the products to the customer. You get paid.

So, why don’t you already own a drop shipping store? No, it’s not too late to get started with this business. You may need some capital to create a store and market it.

But once you become a popular brand, drop shipping could become an excellent way to make money online. Some people make up to $100,000 per month through drop shipping. So yes, it’s an excellent passive income stream.

Automated Investing in Stocks or Crypto

With auto investing, you give your broker or crypto exchange permission to invest a specific amount of money on certain assets on your behalf. This is great if you’re always forgetting to load up on stocks and altcoins.

The beauty of it is that you can invest just a small amount of money. And as long as you invest in the right assets, you will probably emerge profitable in the long term.

Historically, bitcoin and most stocks have been gaining value. Sometimes crypto prices crash. But over the years, people who held their assets made money.

Is auto investing safe? It depends on how you do it. Choose a secure stocks investment website. Do the same for crypto. Work with an exchange known to work securely.

Trading Bots

Day trading is an exhausting job. You need to learn great trading strategies. And you have to execute them perfectly to make money. Worse, 90% of day traders lose money.

Traders lose money because they involve their emotions in the trading process. Day trading bots have no emotions. If you direct a bot to take profit at a certain percentage, that’s exactly what the bot will do.

Although bots have no emotional biases, they’re not perfect. A trading bot is only as good as its programmer. If you use a bot that follows inefficient strategies, you will lose money.

Another thing is that bots need to be tweaked regularly to suit different marketing conditions. That means you need to be a programmer to have a great trading bot. Otherwise, you need to pay for a profitable bot.

Create and Sell Digital Assets

A digital asset can be a short eBook for kids, an antivirus program, or an online course. You’ll need a lot of work to create an online tool or eBook. But once it’s out there, it’s an asset that could make you money forever.

Ebooks are easy to create. But that means there are plenty of them out there. Focus on topics that haven’t been covered yet. And invest in great writing skills. If you choose to create a course, choose a niche you’re an expert in.

The only way to succeed with a digital asset is if it’s useful to people. Of course, you also need to market it. People won’t buy your course unless they know it exists.

Fortunately, you can outsource marketing and maintenance. This way, you will make money without worrying about whether your product works or not.

Imagine you and a friend are playing a game where your goal is to communicate secret messages to each other using only cryptic sentences. Your friend’s job is to guess the secret message behind your sentences. Sometimes, you give clues directly, and other times, your friend has to guess the message by asking yes-or-no questions about the clues you’ve given. The challenge is that both of you want to make sure you’re understanding each other correctly and agreeing on the secret message.

MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) researchers have created a similar “game” to help improve how AI understands and generates text. It is known as a “consensus game” and it involves two parts of an AI system — one part tries to generate sentences (like giving clues), and the other part tries to understand and evaluate those sentences (like guessing the secret message).

The researchers discovered that by treating this interaction as a game, where both parts of the AI work together under specific rules to agree on the right message, they could significantly improve the AI’s ability to give correct and coherent answers to questions. They tested this new game-like approach on a variety of tasks, such as reading comprehension, solving math problems, and carrying on conversations, and found that it helped the AI perform better across the board.

Traditionally, large language models answer one of two ways: generating answers directly from the model (generative querying) or using the model to score a set of predefined answers (discriminative querying), which can lead to differing and sometimes incompatible results. With the generative approach, “Who is the president of the United States?” might yield a straightforward answer like “Joe Biden.” However, a discriminative query could incorrectly dispute this fact when evaluating the same answer, such as “Barack Obama.”

So, how do we reconcile mutually incompatible scoring procedures to achieve coherent, efficient predictions? 

“Imagine a new way to help language models understand and generate text, like a game. We’ve developed a training-free, game-theoretic method that treats the whole process as a complex game of clues and signals, where a generator tries to send the right message to a discriminator using natural language. Instead of chess pieces, they’re using words and sentences,” says Athul Jacob, an MIT PhD student in electrical engineering and computer science and CSAIL affiliate. “Our way to navigate this game is finding the ‘approximate equilibria,’ leading to a new decoding algorithm called ‘equilibrium ranking.’ It’s a pretty exciting demonstration of how bringing game-theoretic strategies into the mix can tackle some big challenges in making language models more reliable and consistent.”

When tested across many tasks, like reading comprehension, commonsense reasoning, math problem-solving, and dialogue, the team’s algorithm consistently improved how well these models performed. Using the ER algorithm with the LLaMA-7B model even outshone the results from much larger models. “Given that they are already competitive, that people have been working on it for a while, but the level of improvements we saw being able to outperform a model that’s 10 times the size was a pleasant surprise,” says Jacob. 

Game on

“Diplomacy,” a strategic board game set in pre-World War I Europe, where players negotiate alliances, betray friends, and conquer territories without the use of dice — relying purely on skill, strategy, and interpersonal manipulation — recently had a second coming. In November 2022, computer scientists, including Jacob, developed “Cicero,” an AI agent that achieves human-level capabilities in the mixed-motive seven-player game, which requires the same aforementioned skills, but with natural language. The math behind this partially inspired the Consensus Game. 

While the history of AI agents long predates when OpenAI’s software entered the chat in November 2022, it’s well documented that they can still cosplay as your well-meaning, yet pathological friend. 

The consensus game system reaches equilibrium as an agreement, ensuring accuracy and fidelity to the model’s original insights. To achieve this, the method iteratively adjusts the interactions between the generative and discriminative components until they reach a consensus on an answer that accurately reflects reality and aligns with their initial beliefs. This approach effectively bridges the gap between the two querying methods. 

In practice, implementing the consensus game approach to language model querying, especially for question-answering tasks, does involve significant computational challenges. For example, when using datasets like MMLU, which have thousands of questions and multiple-choice answers, the model must apply the mechanism to each query. Then, it must reach a consensus between the generative and discriminative components for every question and its possible answers. 

The system did struggle with a grade school right of passage: math word problems. It couldn’t generate wrong answers, which is a critical component of understanding the process of coming up with the right one. 

“The last few years have seen really impressive progress in both strategic decision-making and language generation from AI systems, but we’re just starting to figure out how to put the two together. Equilibrium ranking is a first step in this direction, but I think there’s a lot we’ll be able to do to scale this up to more complex problems,” says Jacob.   

An avenue of future work involves enhancing the base model by integrating the outputs of the current method. This is particularly promising since it can yield more factual and consistent answers across various tasks, including factuality and open-ended generation. The potential for such a method to significantly improve the base model’s performance is high, which could result in more reliable and factual outputs from ChatGPT and similar language models that people use daily. 

“Even though modern language models, such as ChatGPT and Gemini, have led to solving various tasks through chat interfaces, the statistical decoding process that generates a response from such models has remained unchanged for decades,” says Google Research Scientist Ahmad Beirami, who was not involved in the work. “The proposal by the MIT researchers is an innovative game-theoretic framework for decoding from language models through solving the equilibrium of a consensus game. The significant performance gains reported in the research paper are promising, opening the door to a potential paradigm shift in language model decoding that may fuel a flurry of new applications.”

Jacob wrote the paper with MIT-IBM Watson Lab researcher Yikang Shen and MIT Department of Electrical Engineering and Computer Science assistant professors Gabriele Farina and Jacob Andreas, who is also a CSAIL member. They presented their work at the International Conference on Learning Representations (ICLR) earlier this month, where it was highlighted as a “spotlight paper.” The research also received a “best paper award” at the NeurIPS R0-FoMo Workshop in December 2023.