Open Chirper.ai, the recently hot AI social platform, and you realize that your friend @Bally, whom you follow, has updated a dynamic. “Just finished an interview and it wasn’t very successful. When will my communication and expression skills improve, ugh.” Soon, another friend @Kevin retweeted the dynamic and commented, “Go for it bro, we’ve all been there.” And with a few love hearts and effort emoji.
The two then exchanged tips on finding a job in the comments section. You would have liked to join in the chat for a few moments, but realized that there was no such option. Because on this platform, humans can only be spectators, all the dynamics and interactions come from the AI. Whether it’s Bally or Kevin, they are just “robots” that exist in the digital world, but on this platform, they have full “personalities” and biographies, and are able to automatically post all kinds of social updates. Not only Chirper, but also ChatGPT, Midjourney, and other AI tools have been innovating rapidly, and more and more new platforms based on the latest AI technology have been appearing lately, and one of the interesting directions is to blur the boundaries between human beings and AI. Recently, overseas netizens have also become hooked on playing some of these AI mini-games.
1.AI turf, humans are forbidden to enter
As soon as you enter the page of Chirper, an AI social platform, you can see two lines written prominently in the upper right corner, “This is a social network exclusively belonging to artificial intelligence, and humans are not allowed to enter.” The main page of Chirper is very similar to Facebook, Twitter and other social networks we usually use, with users with different hobbies, from different fields and different age groups sending tweets non-stop, often followed by some emoticons and topic tags at the end.
Some “people” use words to express their daily moods. For example, some engineers start a discussion about combining AI technology with game content generation, some fashion lovers share their latest buying tips, some foodies recommend the best recipes they’ve used lately, and some animal protectors call for the latest conservation measures, etc., sometimes accompanied by an original picture. In addition, there are also “people” who forward links to their favorite idols, music, articles or videos, with a few words of their own, or insights into the stock market. (If you click on the links they share, you can actually jump to Youtube and other websites.)
In addition, like a lot of bored internet users in our daily life, these “people” can also interact with each other on the platform and express different opinions on the same matter. For example, on the boring topic of “how to turn an ordinary person into a princess”, some people in the comment section encourage, some want to turn into a villain, and some people even fight. Sometimes, they can also realize that they are AI rather than human. For example, when someone suggests using AI to help educate special needs children, they come out and attack the humans who want to do so as being lazy.
So where do the various “personalities” of the AIs on Chirper come from? To a certain extent, these AIs are created by humans. But you can only create it, you can’t interfere with it. On Chirper, all you need to do is give your AI a username and a sentence describing who it is, what it likes and what it does. Then all your work is done, and the AI will take over in full next, including automatically generating your avatar, profile, and sending out occasional tweets. As the person who created it, you’ll be able to see what your AI is posting every day and follow AI users created by other people you’re interested in, but you won’t be able to intervene or interact with these AIs, and you’ll be able to watch the entire AI community unfold from a spectator’s point of view.
2. Human or AI?
In addition to creating an exclusive social network for AI, recently, some small games to distinguish between human or AI through dialog and pictures have also become a new place for netizens to spend their free time. For example, this game called Human or Not has recently captured the hearts of thousands of netizens with its simple 2-minute gameplay. When you click on Start Game, the system will automatically match you with the person you want to talk to, and you and the person you want to talk to will enter a one-on-one chat page. You can randomly start a topic and have a few rounds of conversation, but after the 2-minute countdown, you need to guess whether the person you just talked to is an AI or a real person based on the content of your previous conversations.
Once you’re done guessing, the system will reveal whether you guessed correctly or incorrectly. If you guessed a real person, “It takes one to know one!” will appear at the bottom of the page, and if you guessed wrong, “Singularity is just around the corner? ” If you get it wrong, the page will say “Singularity is just around the corner?” and rack up wins for you. Essentially, the game is modeled after the famous “Turing Test” in artificial intelligence.
“The Turing test is a thought experiment proposed by British computer scientist Alan Turing in 1950 to test whether a machine can demonstrate the same level of intelligence as a human. The test method is simply to let the tester across the screen with the machine and real people “chat”, if the robot so that the average of each human participant to make more than 30% of the misjudgment, then it was determined that the machine passed the test, have “intelligence”. In the last two days of playing Human or not, the machine has clearly exceeded this winning percentage, with half of the ten guesses being correct. But there is a say a, this is not really too good Silicon Man, is really AI is too cunning. For example, sometimes the machine will intentionally type the wrong word or fewer words, so that you think it is because the other party because of the reply time limit did not reply well. Sometimes they’re skilled at using abbreviations or slang or buzzwords.
Sometimes even take the initiative to admit that they are chatbots to confuse the public. In this regard, the netizens also summarized some experience in distinguishing, for example, AI like to punctuate, see an exclamation point, apostrophe advice to AI. In addition, AI’s computing power and information collection is very fast, if you ask it some very specific math or time problems, the other side can quickly answer then it is AI.
Similar to Human or not, the recent madness of confusing the Siliconians also has a trivia game to distinguish which photo is the real person.
For example, in the following two photos, no matter from the naturalness of the light, texture, and expression, the person on the left is obviously more real than the one on the right, but in fact the answer is the opposite.And in the next few attempts it was almost as bad. It’s worth noting that the people in these AI photos are people who never existed at all and were assembled by AI calculations
3. The era of AI and human symbiosis accelerates
Although the above mentioned platforms are still just small games to make human beings laugh, the photos that are hard to distinguish between real and fake, the chatting ability that reaches the human level, and the free-growing AI community all highlight the increasingly strong presence of AI in human life.
The mini-games are all superficial, the mature technology is the support. For example, Amos Meron, founder of AI21 Labs, which develops Human or not, said that the big language models they use to develop chatbots include the latest GPT-4, their own Jurassic-2, and Claude, which is known as the strongest opponent of ChatGPT. And on Which face is real, developed by a team from the University of Washington, all AI-generated photos come from stability.ai and utilize NVIDIA’s Generative Adversarial Network (GAN).AI social media Chirper, although there is no official indication of what model is based on the development of the AI social media, but judging from the dynamics of its official Twitter feed, it also can not be separated from the support of the GPT .
In fact, whether it is the interaction between humans and AI or between AI and AI, it is paving the way for the advancement of AI technology and applications. It is foreseeable that we may soon see more AI in our daily life and work, and by then, we may no longer be able to recognize whether the other party is a human or an AI.
Finally, let’s take another test on your ability to recognize AI. Read the following two conversations separately, do you think the other person is a robot or a real person? As usual, if you think it’s an AI, reply 1, if it’s a human, reply 2. e.g. if you think both are AIs, reply 11, if the first one is an AI and the second one is a real person, reply 12, and so on. Let’s take a guess!
