HomeTech NewsGamingCicero, the Artificial Intelligence of Meta specialized in negotiation, deception and forecasting

Cicero, the Artificial Intelligence of Meta specialized in negotiation, deception and forecasting

Published on

- Advertisement -

It seems like something out of a science fiction novel, but it is real, there is an AI that is being trained to be an expert in negotiation, deception and foresight.

It is a product that comes out of Meta, Facebook’s parent company. He is a model of artificial intelligence that they have called Cicero, and he already has certain skills of negotiation, deception and foresight, enough to play in diplomacya complex and ruthless strategy game where you have to force alliances, draw up battle plans and negotiate to conquer Europe.

- Advertisement -

While it’s hard to understand why someone might be teaching that kind of thing to a program, it’s a good exercise in tricking humans into thinking it was real. They indicate that it is at a level of sophistication that they thought was years away.

It is capable of retaining information and thinking several steps ahead of opponents, and while it is only used for gaming at the moment, it could in the future be used to come up with more clever scams that extort people, create convincing fakes, or even fall into military hands to design war plans.

It is not the first time that Meta has done something similar. In 2019, Facebook created an AI that could outsmart and beat humans at poker, a much more controversial end than creating art or chatting with humans posing as a person.

Noam Brown, a scientist with Meta’s AI research team, said they are developing the model to impress humanity, to show the limits and advances of artificial intelligence, and they believe Diplomacy is the topic that needs to be worked on now. the world of AI.

How Cicero has been programmed

- Advertisement -

Cicero, or Cicero, has been trained on vast amounts of text data from the Internet and on approximately 50,000 Diplomacy games from webDiplomacy.net, which included transcripts of game discussions.

In the tests they verified that he was sometimes more honest and sometimes more a liar, and that on some occasions he omitted information to win.

If all goes well, this AI will be used to create more believable characters in the metaverse, or more realistic personal assistants.

- Advertisement -

Although the nightmare scenarios are obvious, it is better to be optimistic and think that what is made in Hollywood stays in Hollywood.

You have the details of the good Cicero at ai.facebook.com/research/cicero

- Advertisement -

Latest articles

More like this