fbpx

Can Video Games Influence Artificial Intelligence?

Video Games represent the most realistic learning environment available to people of all ages. Statista reports that individuals across all ages will find themselves playing a video game or two when in one situation or another. Today, you no longer receive the criticism once associated with long hours playing video games, such as laziness, violent tendencies, and being overweight. 

Also, video games have become a source of inspiration and making money. In addition, you can play slots for fun after True Blue casino registration. However, the most amazing fact is that video games aid in artificial intelligence research and their development and usage in a real-life application. Here is how video games can influence artificial intelligence.

  • Training for the real world;
  • Deduction or problem-solving skills;
  • Learning environment through repetition;
  • Lesson transfer.

Training for the Real World

Games, in general, provide the best in terms of intelligence available to human beings today! At the same time, they also provide the best testing mechanisms for working and practical artificial intelligence algorithms. AI researchers have been entering AI algorithms into contests and having the algorithm play against gamers as a learning process. Professor Julian Togelius from the Tandon School of Engineering at the New York University indicates that “games provide good testing authority as the game design and gameplay revolves around ensuring the gradual progression of the human mind through the provision of complex challenges as one progress higher in the gaming levels.”

However, some of the features retain their shape, surrounding remains the same, but the tasks do change. An AI can adapt to the changing tasks and learn from the repetitive environment. For instance, Artur Filipowicz uses the “Grand Theft Auto V” video game as a learning tool for his AI software. The aim is to help the software learn the different stop signs which vary depending on the country, region, or continent, and personality variations. The game provides life-like representations of the stop sign, and the software stops at every sign, indicating a successful learning process.

Like Grand Theft Auto V, AI researchers can utilize video games as real-life applications or simulations for AI. However, creating a real-life environment is not only expensive; it will also come with extensive repercussions, which adds to the suitability of using video games.

Deduction or Problem-solving Skills

Video games challenge every level of the human brain. The designers employ a game design that guides players to focus on specific elements and find solutions on how best to overcome a hurdle or challenge. A major factor that many people would not identify with is that an individual is constantly learning and adapting to suit his goals through gaming. AI researchers desire to have a similar working framework where the software would focus on a target, evaluate the available means of getting to it and make decisions and deductions on the most strategic routes to take while in the process! 

In that way, video games can pick complex challenges, shred them into smaller but manageable areas and finally, provide workable solutions within a specific timeframe. At the same time, researchers get to study their software and machinery at work (Robots). The researchers can deduce the machine’s working pattern and process of understanding its current environment or depending on the assigned task.

For instance, DeepMind has been learning how to play Atari games and has recently become a master. The AI software by Google beat the world’s Go Champion, a feat that was once considered impossible. The developers indicate that the AI has been learning on its own and managed the feat without any assistance. The only challenge is that the game considers scores to measure true success but does not define ‘the how” to the AI.

Learning Environment Through Repetition

Using robots as learning tools has become moot since the software can execute tasks almost 1000 times faster than a robot. Therefore, many AI researchers use artificial intelligence software to execute tasks since they do not reheat and won’t cost a “dollar more” as it also doesn’t break. For instance, the OpenAI Universe recently demonstrated an algorithm laughing how to open and manipulate a cube for over a century (100 years). The algorithm employed a trial and error approach, with learning outcomes on record and changes to their approaches recorded. It showed that AI could indeed learn from repetitive actions!

In other words, video games manipulation can provide faster and infinitely repeatable, and realistic learning environments for artificial intelligence. For instance, DeepMind picked and perfected “Tunnelling” after playing 600 times and beating “Breakout” in the process. A person can carry out out the task and try learning the game for days or weeks to come, but for an AI, a few hours are enough for a successful learning process.

Lesson Transfer

Even with Repetition offering an executable and understandable learning process for machine intelligence, the gap between machines and human intelligence is still wide. For example, you can define what top casino bonuses are to an AI. Still, you cannot define how to manipulate the bonuses given or take advantage of any opportunities to help meet all set rules by the casino before accessing any associated bonus wins! If it executes simple, straightforward tasks like “Repeat” repeatedly in a series of experiments, then an AI beats the human brain. 

However, in the prevalence of twists and turns like in the bonus example where repetition is present but for different factors or elements all contributing towards one major goal, AI cannot survive against the human brain.

Researchers are trying to transfer the ability to hold complex concepts and apply them to different and unrelated environments and achieving positive results or outcomes. The aim is to train the AI to pick lessons learned from the continuous and repetitive actions when playing a game like Breakout and use the same principle to play another game genre or title like Portal 2. 

For instance, associate professor Togelius dictates that training an AI in StarCraft, one of the complex online video games offering a multiplayer option, requires players to employ quality management skills and strategic planning to overcome all present challenges gameplay. He says that “you are making tactical plans, appoint positions strategically, bring together a working team while building and coordinating all associated logistical functions. At the same time, a player has to ensure that players receive all resources available in a seamless manner.” He concludes, “If a human mind can learn and develop, then an AI can too.”

Summary

Yes, video games can influence artificial intelligence resulting in smarter and more capable software. First, video games can help train AI to spot real and actual elements in the real world, such as stop signs and children crossing the road. Second, it can help develop AI problem-solving skills through deductions and problem identification. Third, the AI can break the problem down and develop a working solution or alternative to it. 

Finally, AI can transfer lessons after carrying out repetitive tasks to a different environment and a completely varying problem or task! In other words, video games are successfully changing the learning process of AI today. As a result, researchers can develop AI, and the outcomes would be positive and more valid impact!

Ajay Deep

Ajay Deep is the brain behind Coworking Mag. He founded this website to help startups and aspiring entrepreneurs find a coworking space in their city. He is a successful entrepreneur who started and scaled a bunch of startups – all from shared office spaces. He has visited hundreds of coworking spaces in different countries and is now an investor in this evergrowing idea of developing new coworking spaces. You may reach Ajay Deep at hello@coworkingmag.com
Back to top button