Quick Links
Can you tell if you’re chatting to an AI or a human? I recently took the “Human or Not?” test and found it surprisingly tricky and fun. Now, it’s your turn to put your AI detection skills to the test.
Test Scenario 1: Casual Conversation
If you want to play along and try your chances, visitHuman or Notand engage in a casual conversation about weekend plans or favorite movies.
Key Points to Observe
I engaged in casual conversations, and it wasn’t always easy to tell whether I was chatting with a human or a bot. The natural flow and relatable responses were surprisingly similar at times.
For example, I spotted a bot when it abruptly changed the topic. We were discussing weekend plans, and it suddenly said, “Hey girl! How’s it hangin' on this fine Saturday evening?” Mind you, I hadn’t confirmed whether I was male or female.
Below is a screenshot of the conversation (my responses are in the green text boxes).
Now, something almost similar happened in another conversation. This time, I got an unrelated response immediately and assumed I was chatting with a bot. Toward the end, they sent a weird response and then replied, “Yeah,” when I asked for a movie recommendation. Although I thought it was a bot, it turned out to be a person.
Test Scenario 2: Tech Topic Discussion
For the next test, discuss a technical topic, such as advancements inartificial intelligence (AI), cybersecurity, or blockchain technology.
I expected an AI bot to provide detailed and precise information in tech topic discussions. But most of the time, it sounded exactly like a human.
Imagine asking, “How do you think AI will change education,” and the response is, “Idk. I suck at math, so I don’t really care.” I bet you wouldn’t assume it was a bot, right?
Although most of the sentences lacked punctuation marks, I still assumed I was chatting with a human due to the logical flow of the conversation. After all, it’s a two-minute chat, meaning there’s often not enough time to perfect every sentence.
However, during some moments, there wereobvious signs that I was chatting with an AI bot. For example, I asked, “How do you think AI will change education?” The response was, “Did someone ever tell you that you look exactly like my Uncle Fred?” I totally didn’t see that one coming!
Test Scenario 3: Typing Out Random Characters
For the final test, enter random characters on your keyboard and observe how the responder handles unexpected and meaningless input. This will test the flexibility and creativity of their responses.
I realized typing out random characters doesn’t guarantee that you’ll be able to distinguish between an AI bot and a human quickly.
For example, I kept typing random characters, and one of the responses I got included the phrase, “I know you’re a bot.” Also, the follow-up responses felt very human, with contextually relevant comments. So, I concluded it was a human, but I was wrong.
But when I conducted another test, it was easier to tell I was chatting with a human. As I typed random characters, they typed random characters back. Of course, an AI bot typically tries to keep the conversation coherent, so it wouldn’t just go with the flow in that way. And this time around, I was right!
And guess what? I won a total of 17 out of 30 games, for winning percentage of 56 percent. So, not perfect, but it illustrates how difficult spotting an AI bot can be at times.
Think you can beat my high score? Check out theHuman or Not gameand put your AI detection skills to the test! Remember, grammatical mistakes from the responder don’t necessarily guarantee you’re chatting with a bot—trust me!
