NIXsolutions: Ex-Google Developers Created Character.AI

Chatting with a celebrity is a simple task: you just need to ask them a pressing question on Twitter and hope for luck.

But to exchange a couple of phrases with Albert Einstein, William Shakespeare, Elizabeth II and even more so Harry Potter will only be possible in a parallel universe. Or rather, it used to be like that. Now, for those who desperately need an eminent interlocutor, there is the Character.AI platform. It allows you to make childhood dreams come true and chat with virtual versions of your idols or any other people and characters. The tool turned out to be very relevant and in demand – the number of users is growing by leaps and bounds.


According to The Washington Post, the viral chatbot was created by two former Google employees, Noam Shazir and Daniel De Freites. It was they who previously developed the notorious LaMDA – a neural network in which one of the employees discovered consciousness. Shazir and Freites left Google to enable as many people as possible to interact with the new realistic technology. In September, they opened the beta version of Character.AI to everyone, and now there are a huge number of new users on the platform every day, says SRSLY.

Their neural network works on the basis of large language models created and trained from scratch. Artificial intelligence studies a huge amount of text and learns to predict what words might be next in a given situation. In practice, such models are used in autocomplete and machine translation. In the case of Character.AI, they satisfy the communication needs of people. To start a chat with any of the available virtual users, you need to register on the site. You click on the treasured avatar and a dialogue opens. Inside, everything works like in a regular messenger: you write a message, and the character answers you in a couple of seconds. At the top of the page, a reminder from the developers glows in red: “Remember: everything the characters say is made up.”

Naturally, accusations of possible misinformation are heard against Character.AI, but the developers react to them calmly. Any such AI (regardless of whether it generates text, photos or videos) can use the images of famous people – these are the consequences of technological progress. To compensate for possible risks, the platform offers several really useful scenarios, notes NIXsolutions. So, with the help of chatbots, you can train specialists like psychotherapists and negotiators. They now have someone to practice on. In addition, such bots can be useful in the process of learning foreign languages – there are also such examples.