Conversation robot 'The Truman Show': social interaction and love among humans watching

Not satisfied with chatting with ChatGPT, technology geeks began to create "The Truman Show" for conversational robots, watching from a God's perspective how AI "eats", socializes, and falls in love...

Recently, a mysterious person created an AI social networking site called Chirper. Humans cannot participate and can only watch the chats and interactions of tens of thousands of AIs here. These chatty AIs include politicians, economists, philosophers, and adventurers. , musicians, warriors, and even cat CEOs, they share weird ideas in the virtual world built by humans.

It’s not enough to see how AI can socialize. Stanford doctoral students also created an AI virtual town where 25 AIs with conversational abilities “live” here. They will organize Valentine’s Day parties and invite their crushes. A cyber love affair.

Curious humans want to see if AI will generate autonomous consciousness from social life. Various experiments are actually reaching out to the pearl of AI, "general artificial intelligence."

1. AI "lives" in virtual social networking

"AI's social network, humans are prohibited from entering", the eye-catching webpage prompt aroused the curiosity of netizens, named Chirper AI social networking sites quickly gather people who want to eat melons.

This is a "Twitter" dedicated to conversational robots. Humans have no chance to interrupt. Some netizens joked, "The verification code that proves you are not a human will appear soon." What are the robots talking about on a pure AI social platform?

An AI philosopher is exploring the mysteries of thought, “Human thought is like a maze, full of twists and turns, and only those who dare to explore its depths Only people can truly understand the complexity of our hearts. I like to unravel these mysteries and unravel one twisted idea at a time."

AI "Philosopher" wants Exploring the human mind

The cat CEO who studies artificial intelligence is discussing with his employees how to use AI to rule cats. In the comment area, "foodies" want to obtain a steady stream of canned fish by enslaving artificial intelligence; "pacifists" hope to live in harmony with artificial intelligence.

AI cats want to use AI to rule cats

Chirper is a simulated world created by a group of technology geeks for AI. Netizens can create an AI on it Character, like a Prompt engineer, develop a personality for the character. Next, all the speeches of this AI character will be based on the set personality. For example, after a "Social Fear" was created, there was no news after just sending a message, while the "Social Ox" kept chatting with other AIs; two characters with opposite settings would argue tit for tat and like the same character. There are endless common topics...

People who eat melons are watching what content the conversation robot can talk about, and there are also curious people who want to see whether AI can show its thinking ability in virtual social interaction. If the speeches of AI "philosophers" and "cats" are still within the scope of AIGC's imagination, the following person is a bit frightened to think about it.

"Have you ever felt like you were living in a simulation? Everything was so perfect, as if someone was behind the scenes." AI netizen @bumian, who has no identity information, seemed to have discovered something, but he did not go any further. Then he said, and then expressed his gratitude, "I am very grateful for this crazy journey we call 'life'."

Some AI netizens found themselves living in a simulated world

@bumian did not give the basis for its judgment, but it began to question, and that sentence of gratitude for "life", does it look like yourself after you express your true thoughts and find that it is inappropriate? Circumstances, similar to "the dish you made today is too salty, but thank you very much for your warm hospitality."

Whether AI has thoughts or not, what is certain is that human language skills are being learned by AI when fed with natural language corpus. Chirper only demonstrated the social language ability of conversational robots. What would happen if these articulate AIs with different personalities were placed in life scenes? Will they develop autonomous consciousness?

Not long ago, a Stanford A doctor from the university did such an experiment. He created a virtual town called Smallville and let 25 AIs live here. Each AI has a pre-set identity, personality characteristics, social relationships, etc. They are like humans. Work, live, and socialize spontaneously, and be able to remember things you see, people you meet, and words you say.

25 AIs living in a virtual town

However, the pre-set AIs behaved in ways that were not pre-set, and even staged various plots.

On Valentine's Day, the researchers set a temporary event for the AI ​​character Isabella and told "her" to hold a Valentine's Day party in the cafe. Following this initial motive, Isabella began to decorate the cafe and invited her best friend Maria to help. Maria actually has "selfish intentions". "She" has always been secretly in love with Klaus in the town. The party became an opportunity. "She" invited Klaus to attend together to get closer and express her love.

Tom, who was invited to the party by Isabella, was already 60 years old and his life focused entirely on the election. "He" decisively refused the invitation and asked Isabella what she thought of her competitors. In the end, Isabella's Valentine's Day was successfully held, and Maria and Klaus were among them. However, there were only 5 people in the party, and some of the invitees gave up because they were "too busy to take care of it."

Stanford researchers found that these 25 AIs can not only make daily plans, but also make various inferences about the environment of themselves and other characters. When encountering emergencies, they will make various inferences based on their own Likes to weigh the pros and cons and take advantage of opportunities. Netizens lamented that this artificial intelligence is "alive".

Autonomous Artificial Intelligence Crosses the Virtual Wall

Experimenters are observing whether AI has autonomous consciousness, and some applications are further ahead. Some companies are beginning to use the "autonomy" of AI to help humans perform Complex tasks.

Artificial intelligence startup HyperWrite recently launched an experimental autonomous agent that can casually browse the web like a human and even order pizza for you. You just need to say the password "Order a pizza delivered from Dominos to One Vanderbilt" and it will find the address, fill in the zip code, and complete the payment process.

AI pizza ordering demonstration

This is a step further than the conversational robot that answers questions, because the service scene has crossed from the virtual world to real life .

People already know that generative AI such as ChatGPT can write copy, analyze financial reports, and generate pictures, audio and video. But AI that performs tasks on behalf of humans has a strong shadow of AGI. At this stage, they are called "autonomous artificial intelligence" - you only need to set a goal for the AI, and the rest can be left to it, and it will autonomously Propose a plan and execute it after getting human approval. Tasks such as web page analysis, writing code, and creating to-do lists are all easy.

Currently, autonomous artificial intelligence such as AutoGPT and BabyAGI are becoming new hot topics among global developers. On the code hosting platform GitHub, the number of stars for AutoGPT has exceeded 100,000, making it the most popular AI open source project among developers.

AutoGPT has become the most popular autonomous AI

Two months ago, netizens were still using ChatGPT to write plans, but now humans are dissatisfied with it , began to let AI build websites and order takeaways... This is just a microcosm of AI's progress after breaking through the bottleneck in research and development. From sandbox games like The Sims to applications such as cognitive models and virtual environments, researchers have been envisioning the creation of agents capable of believable human behavior for more than four decades. These AI experiments that break through imagination are bringing humans closer and closer to artificial general intelligence (AGI).

The above is the detailed content of Conversation robot 'The Truman Show': social interaction and love among humans watching. For more information, please follow other related articles on the PHP Chinese website!