Human or Not? (AI Game)

What is Human or Not?

“Human or Not?” was an online game developed by AI21 Labs, positioning itself as a “social Turing game.”

Advertisements

Drawing inspiration from the Turing Test formulated by mathematician Alan Turing, the game aimed to evaluate if machines can exhibit intelligent behavior indistinguishable from humans.

Participants engaged in two-minute text chats with an unknown entity, which could be either a human or an AI bot, with the goal of determining the nature of their conversation partner.

The game employed a variety of Large Language Models (LLMs), including GPT-4, Claude, Cohere, and AI21 Labs’s Jurassic-2, showcasing the latest in natural language processing technology.

How to Play Human or Not?

Engaging in “Human or Not?” was straightforward, upon visiting the game’s dedicated website, players could start a game with a single click.

The system then randomly matched them with another player or an AI bot, also randomly assigning the conversation’s initiator.

Each participant had a 20-second timer for their messages; if time ran out, the message was sent as-is, and if no message was written, the conversation prematurely ended.

Following the two-minute dialogue, players must decide if they are interacting with a human or an AI, with immediate feedback provided on the accuracy of their guess.

While the game was open to everyone, it was not intended for children under 18, underscoring AI21 Labs‘ commitment to responsible gaming.

Why is Human or Not Popular?

The popularity of “Human or Not?” can be attributed to various factors, firstly, the game taps into the public’s growing curiosity and debate surrounding AI and its capabilities, providing a direct, interactive platform for exploration.

Over 2 million participants have engaged in the game, drawn by the challenge of distinguishing between humans and AI.

The game’s results, with nearly one-third of participants guessing incorrectly, underscore the sophistication of modern AI, fostering discussions about AI’s role in society and its ethical implications.

Additionally, the social experiment aspect of the game adds an intriguing layer, making participation feel meaningful and contributing to a larger understanding of AI.

What Was the Outcome of the Human or Not Experiment?

Following the closure of the “Human or Not?” experiment for the foreseeable future, there has been a huge amount of data produced by more than 15 million conversations conducted in the Turing test game.

So far, AI21 Studios has published the results from the first two million conversations and guesses reviewed.

Overall, it was found that 68% of people guessed correctly when asked to determine whether they were interacting with an AI chatbot.

However, it was also found that people found it easier to identify a fellow human in 73% of cases studied.

In terms of geographical performance, France scored highest on correct guesses, scoring 71.3%, compared to a global average of 68%. While India was found to have the lowest scores at 63.5%.

Human or Not - Correct Guesses by Country

As for demographics, women outperformed men by a slim majority, while younger users tended to outperform older users by a slight degree.

The Bottom Line

In conclusion, “Human or Not?” serves as a fascinating reflection of AI’s current state and its interaction with society.

By providing a direct, interactive platform for the public to engage with AI, AI21 Labs has not only created a viral sensation but has also contributed to the ongoing discourse on AI’s capabilities and ethical considerations.

The game’s results affirm Alan Turing’s predictions, demonstrating that we are at a point where AI can convincingly mimic human conversation, challenging our assumptions and prompting critical reflections on how we envision AI’s role in our future.

Whether for curiosity, debate, or scientific inquiry, “Human or Not?” has established itself as a pivotal tool in understanding and navigating the complex world of AI and its societal implications.

Advertisements

Related Terms

Latest Artificial Intelligence Terms

Related Reading

Sam Cooling

Sam is a technology journalist with a focus on cryptocurrency and AI market news, based in London – his work has been published in Yahoo News, Yahoo Finance, Coin Rivet, CryptoNews.com, Business2Community, and Techopedia. With a Master’s Degree in Development Management from the London School of Economics, Sam has previously worked as a Data Technology Consultant for The Fairtrade Foundation and as a Junior Research Fellow for the Defence Academy of the UK. He has traded cryptocurrency actively since 2020, actively contributing to Fetch.ai and Landshare.io. Sam’s passion for the crypto space is fuelled by the potential of decentralisation technology…