Meta's AI just became the first to outwit human players in Diplomacy

Is there any strategy or alliance too complex for its ruthless intelligence?
Kavita Verma
Meta head office
Meta head office

Derick Hudson/iStock 

Meta AI made a groundbreaking announcement with the launch of CICERO - the first-ever AI (artificial intelligence) capable of winning at Diplomacy, a multiplayer strategy game that calls for mutual trust, compromise, and teamwork. This marks a significant milestone in AI evolution.

Meta stated, 'AI has come a long way, demonstrating its capabilities by beating grandmasters at chess, Alpha Go, and even poker.' With CICERO, they have achieved building something that can apply groundbreaking intelligence across different tasks.

"CICERO moves AI from the gaming table into real-world applications." 

What is Meta AI's CICERO?

CICERO is a revolutionary technology that uses strategic reasoning and sophisticated dialogue modeling to create conversations that sound natural but have an underlying plan.

From the moment each game starts, CICERO assesses its environment and maps out potential courses of action. Apparently, it does this all while generating realistic interactions with other players in order to achieve maximum success.

At its core, CICERO employs a tremendous 'controllable dialogue model,' which allows it to converse as naturally and fluently as if one were talking with an actual person. 

To develop this inimitable skill set, Meta's team began by training the machine on 2.7 billion parameters based on text scraped from across the Internet before tinkering further and fine-tuning over 40 thousand human interactions taken from webDiplomacy.net.  

CICERO (AI) outwits humans in the classic game of Diplomacy

Meta set out to develop CICERO - a computer agent with an impressive knack for negotiation. 

After playing countless rounds on webDiplomacy.net, CICERO's strategic calculations put him ahead of his human competitors. In fact, he earned more than double the average score among players who had completed multiple games.

By combining powerful AI models for strategic thinking and natural language processing, Meta CICERO can outsmart any other virtual or human player. 

Meta's future plans with CICERO

Meta stated that their CICERO research would help ease communications between AI and humans. For instance, this technology could guide people in learning a new skill or even power a video game where the AI players converse like humans. Also, they will understand the player's skills and motivations and adapt accordingly. 

While it's a wonderful innovation, it has some risks too. It can be used to manipulate humans by impersonating people and misleading them, which can be dangerous depending on the context. 

While the risk cannot be eliminated entirely, Meta has committed itself to detecting and blocking 'toxic messages' which may arise from online texts ingested while training its system. 

The hope is that other researchers can pick up where they left off in a responsible manner so these advances won't go to waste.

Add Interesting Engineering to your Google News feed.
Add Interesting Engineering to your Google News feed.
message circleSHOW COMMENT (1)chevron
Job Board