Skip to main content

Artificial intelligence beats 'Ms. Pac-Man' with a perfect score of 990,990

Divide and conquer: How Microsoft researchers used AI to master Ms. Pac-Man
A team of researchers have managed to develop an artificial intelligence capable of mastering the arcade classic Ms. Pac-Man. Maluuba — a Canadian deep learning startup that was acquired by Microsoft in January 2017 — used a divide-and-conquer technique to empower its system to complete the Atari 2600 version of the game with a perfect score of 999,990.

Maluuba’s approach is interesting, because it breaks down the strategies and maneuvers required to beat the game into their component parts. Various different agents focus on one job and one job alone, while an agent put in charge of managing from the top makes high-level decisions about what actions should be prioritized.

Recommended Videos

For instance, some agents might be tasked with chasing down pellets, while others focused on avoiding enemies. The decision-making agent would then choose the best option based on weighted logic — if a hundred agents wanted to move left to grab a pellet, but only three wanted to move right to avoid a ghost, it would elect to move right because colliding with the enemy would end the run.

Please enable Javascript to view this content

Ms. Pac-Man is relatively widely used in AI research because of the unpredictable nature of its gameplay, according to a post on the official Microsoft blog. Steve Golson, who is credited as co-creator of the original arcade version of the game, notes that this was intentional, as the game was reliant on players spending quarter after quarter on extra lives for it to be a financial success.

Maluuba used reinforcement learning, a process by which an AI receives positive or negative feedback for each attempt it makes at a problem, to address this unpredictability. It’s hoped that reinforcement learning could help foster systems that are better equipped to make decisions on their own, compared to those that are trained via supervised learning, where the system is simply fed good and bad examples to establish a base of experience.

Brad Jones
Former Digital Trends Contributor
Brad is an English-born writer currently splitting his time between Edinburgh and Pennsylvania. You can find him on Twitter…
Researchers use artificial intelligence to develop powerful new antibiotic
MIT researchers used a machine-learning algorithm to identify a drug called halicin that kills many strains of bacteria. Halicin (top row) prevented the development of antibiotic resistance in E. coli, while ciprofloxacin (bottom row) did not.

 

Researchers at MIT have used artificial intelligence to develop a new antibiotic compound that can kill even some antibiotic-resistant strains of bacteria. They created a computer model of millions of chemical compounds and used a machine-learning algorithm to pick out those which could be effective antibiotics, then selected one particular compound for testing and found it to be effective against E. coli and other bacteria in mouse models.

Read more
MIT’s latest artificial intelligence can rewrite outdated Wikipedia pages
wikipedia asiacell iraq mwc2017 politics

 

Writers, editors, and researchers: Hold on to your red pens. Because MIT-powered A.I. may be coming for your jobs. 

Read more
Turns out, it’s not that hard to do what OpenAI does for less
OpenAI's new typeface OpenAI Sans

Even as OpenAI continues clinging to its assertion that the only path to AGI lies through massive financial and energy expenditures, independent researchers are leveraging open-source technologies to match the performance of its most powerful models -- and do so at a fraction of the price.

Last Friday, a unified team from Stanford University and the University of Washington announced that they had trained a math and coding-focused large language model that performs as well as OpenAI's o1 and DeepSeek's R1 reasoning models. It cost just $50 in cloud compute credits to build. The team reportedly used an off-the-shelf base model, then distilled Google's Gemini 2.0 Flash Thinking Experimental model into it. The process of distilling AIs involves pulling the relevant information to complete a specific task from a larger AI model and transferring it to a smaller one.

Read more