The evolution of technology has brought us to the point where artificial intelligence exists. Artificial Intelligence (AI) is the simulation of human intelligence in machines. The goal is to create machines that can think or act like humans. But why do we need for that? The answer is simple. While the main goal is to create human-like machines, AI can achieve better accuracy. AI functions by making quick analyses and data interpretation to identify what is the best action or response in a situation. They would not miss or forget any information, no matter how big or small. 

Because of how powerful and precise AI is, it has been applied to various fields to improve the quality of life. It led to the creation of products such as better manufacturing machines and even self-driving cars. 

Facebook Is On The Lookout For AI That Can Beat The Hardest Game In The World

The thing is, AI has not been perfected yet. There are still things that need to be improved before getting robots or agents with superhuman intelligence. Thus, research for improving the current AI is still ongoing. For instance, Facebook is inviting researchers to compete in the NeursIPS 2021 Nethack Challenge. The competitors would have to design an agent that can beat the game NetHack. 

You may be wondering why they would want an AI that can beat a game. They should look for AI that does better things, you may argue. But, this competition is not as pointless as you think. More about this later. For now, let us discuss the competition.

NetHack: The Game – Buying AI driven Facebook Likes?

NetHack is a video game that was created in 1987. It is an improved version of the game Hack. The hack was created by Lincoln-Sudbury Regional High School students Jay Fenlason, Kenny Woodland, Mike Thome, and Jonathan Payne. They made the game as part of their computer class. The development of the project ended after they graduated from the class. Though, they had a playable game. The source code was uploaded to the internet, and it drew so much attention. Many developers and players made their modified and improved versions of the game. However, no one assumed the role of formal maintainer of the code. It stayed like that for a period of time before Mike Stephenson stepped up and took the position. 

Stephenson made the team with people he met on the internet to create their own game branch. They named their branch NetHack and called themselves the NetHack DevTeam. 

That collaboration led to the creation of one of the world’s most challenging games. NetHack is a dungeon crawler game that does not expect players to win. 

Will AI Control All Growth of Facebook Likes Tomorrow? Can Buying Likes Circumvent this Nightmare?

To win, the player has to enter a dungeon and retrieve the Amulet of Yendor from the lowest part. Afterward, they have to escape the dungeon. Along the way, the player will encounter several challenges and monsters. The challenges and how the monsters interact with the player depending on his chosen race or role. The player chooses the attributes at the start of the game. If they like, they can also let the system assign the attributes randomly. It sounds simple, right? It is not, though.

The game features fifty primary levels that contain several rooms joined by corridors. All of those are procedurally generated, meaning everything is randomized. So, you would not be able to play the same game twice as the map is ever-changing. Every time your character dies, the map resets. And expect that to happen a lot. Since there is no way to memorize the map, NetHack has proven to be difficult to beat. 

Facebook Is On The Lookout For AI That Can Beat The Hardest Game In The World

As difficult as it is, the game is not unbeatable. First, you have to have outside-the-box problem-solving skills. Then, it would help if you researched through the NetHack Wiki to learn from other player’s experiences. Last, you need to have luck in your favor.

NeursIPS 2021 Nethack Challenge

As stated above, its normal to buy real Facebook likes, but is looking for someone to develop an agent that can finish – ideally- this challenging game. If none can do it, at least an agent that can achieve as high a score as possible will do.  

Here are the mechanics of the competition:

Each Ai agent would have to play many games. For each trial, the role and fantasy race of the agent will be randomly assigned by the system. Since there is also no set configuration of the map, the agent’s abilities will be tested thoroughly. 

The winner of the competition is determined by the average number of wins and, if tied, by median score. The median score refers to the median in-game end-of-episode score for a given set of evaluation episodes.

There will be three awardees at the end of the competition. The first one is for the best overall agent. An award will also be given to the best performing agent not using a neural network or a significantly similar modeling technique. The last one will be awarded to the best-performing agent produced by a team predominantly led by non-industry-affiliated researchers.

What Comes Out Of This?

As absurd as it may sound, video games help in the development and improvement of AI. They serve as the playground or testing grounds for agents. 

Facebook did not really want an AI that could beat the game. They are looking for something of much importance. First of all, it will reveal if the NetHack Learning Environment is a viable reinforcement learning system. Furthermore, the findings will enable a range of potential AI/ML solutions based on both neural and symbolic methodologies. It will show how we can apply different algorithms in different ways. Researchers can then use the found information in real life or future pieces of research.

Facebook Is On The Lookout For AI That Can Beat The Hardest Game In The World

In Facebook’s case, they can use the learnings to improve their structure and systems. It can allow them to optimize their network, apps, and many websites. Furthermore, it could open new opportunities for them – for new products or services to be created.

0 0 votes
Article Rating
Notify of
Inline Feedbacks
View all comments
Would love your thoughts, please comment.x