Facebook's CherryPi suffered defeat in 'StarCraft,' but a win for its AI team

By Kevin Parrish Published October 10, 2017

Even the brilliant scientists at Facebook can’t seem to defeat the mighty ZZZKBot. The company participated in a recent StarCraft tournament that pits artificial intelligent “bots” against each other instead of humans. Facebook’s “CherryPi” AI bot walked away with a lower lip, having ranked a mere sixth place in the overall tournament with 2,049 wins out of 2,966 games.

The StarCraft AI competition took place during AIIDE (Artificial Intelligence and Interactive Digital Entertainment) 2017 at the Snowbird Ski Resort in Utah from October 5 to October 9. The annual conference focuses on artificial intelligence and interactive entertainment, pooling talents from the commercial and research communities. The highlighted topic covered artificial intelligence beyond its use in gaming.

Recommended Videos

The StarCraft AI competition specifically used the original StarCraft PC game and its expansion, Brood War. Its purpose was to evaluate the state of artificial intelligence, and how it handles real-time strategy games. This genre poses a challenge because the player must scan the field, manage resources, investigate unknown environments, and respond quickly to threats. Chess doesn’t offer that strategic depth.

Although StarCraft: Brood War has an integrated artificial opponent, it has access to the game’s back end. It doesn’t view the screen as a human does — it’s not a detached entity that must think on its own. Thus, for an external “bot” to experience the game like a human, a special interface called BWAPI was created for bots to interact with the PC game.

“BWAPI only reveals the visible parts of the game state to AI modules by default,” the Github notes state. “Information on units that have gone back into the fog of war is denied to the AI. This enables programmers to write competitive non-cheating AIs that must plan and operate under partial information conditions.”

In Facebook’s defense, the ZZZKBot is a mean opponent. It just participated in the 2017 IEEE SIG StarCraft AI competition in August and came out winning 1,790 games out of 2,374. At the end of the local tournament in Utah, the ZZZKBot came out on top again with 2,465 wins out of 2,966 games. Written by Chris Coxe, it first made an appearance during the AIIDE 2015 conference.

But given that Facebook’s CherryPi bot managed sixth place out of 28 opponents in its first swing, the company did not exactly suffer total defeat. Facebook’s AI team behind CherryPi are composed of eight individuals who published a dataset last week drawing on early data collected from CherryPi’s gameplay. The data includes 496 million player actions and millions of captured frames.

All bots were ranked by their final winning percentage of one-on-one games. They could not cheat and could not take advantage of in-game glitches. Each session lasted up to 60 minutes with fog of war enabled, which covers unexplored areas of the field. Bots were penalized for slow computations.

Out of the 28 competing bots, 15 were created and submitted by independent developers. Even more, all five bots ranking higher than CherryPI were created by independent developers. The only other non-independent AI developer to reach the top 10 was Stanford University and its Arrakhammer bot.

Former Contributor

Kevin started taking PCs apart in the 90s when Quake was on the way and his PC lacked the required components. Since then…

Topics

Computing

Look, the MacBook Neo is a fine machine for building apps. And yes, I am utterly surprised

I am talking about apps with a generous dose of AI wizardry happening in the background.

MacBook Neo in yellow.

In a span of just one week, I have built a full-fledged Android app for my Galaxy Z Fold 8 foldable phone, three Mac apps, and an iPhone app. And here's the fun part. The entire journey of creating these apps actually started on a phone, jumped to my MacBook Neo, compiled just fine, and ran flawlessly, without making me want to tear my hair off. The latest example?

It built my calendar-clipboard hybrid app in Terminal within 20 seconds, and an on-device transcription app with a retro look in roughly 15 seconds. I used to think building desktop software would have me burning the midnight oil. The MacBook Neo did it in less time than I can pour myself a can of soda and take the first sip.

Computing

AI could be doing to our tolerance for hard work what social media did to our patience

Even brief AI use may make unaided work feel more frustrating once the help disappears.

ChatGPT on iPhone

It is no secret that social media, and especially the rise of short-form content, has played a major role in the declining patience many of us have for anything that takes too long to get to the point. A few seconds without something interesting happening is enough to make us swipe away and look for the next thing. AI appears to be doing something similar to the way we approach work.

A new peer-reviewed study found that people who used AI for around 10 to 15 minutes were more likely to struggle and give up once the AI was taken away. The researchers ran three randomized experiments with 1,222 participants who were given math and reading-comprehension tasks. AI helped them perform better at first, but their performance tanked once they had to solve the problems on their own.

Computing

Microsoft wants Copilot Notebooks to handle the work that comes after an AI answer

Built-in Office editing gives users somewhere to revise what Copilot produces

Copilot graphics

Microsoft is bringing Office editing into Copilot Notebooks through its Frontier preview. Users can work on Word, Excel and PowerPoint files alongside their project references, making changes themselves or asking Copilot to handle them.

Generating a report or presentation with AI still leaves someone responsible for getting it ready to share. The October update puts editing tools where that work begins, though having a document neatly formatted doesn’t make its contents correct.