Yandex voice assistant not only helps users in everyday communication with equipment, but also entertains. Jokes with Alice from Yandex have become an integral part of collections of Internet memes, so let's laugh too.
Yandex's voice assistant is a serious application that uses a multilayer neural network for training. Despite the fact that it all sounds boring and clever, Alice jokes well and understands jokes herself. To bring the assistant's reaction closer to the natural one, the developers provided the voice assistant with an excellent sense of humor and jokes with alice from Yandex flooded the Internet. Since the program appeared in October 2017, hundreds of funny replies have appeared on the network, which have become memes. And trolling of alice yandex has become a kind of sports competition.
How to mentally traumatize a voice assistant
Yandex immediately explained that this is not a primitive chat bot with a speech recognition function, but a full-fledged interlocutor. With its own character, musical preferences and other features that distinguish a software bot from a living person. Yandex began to spread the first jokes about Alice itself in order to attract as many people as possible to use the new program.
As was the case with other virtual assistants (Siri, Cortrana, Google Assistant, Alexa), users immediately wanted to test the newcomer for strength, trying to offend or offend her. It quickly became clear that the developers put stubbornness, malice, even to some extent impudence and, of course, a sense of humor into the character of the virtual assistant. Yandex Alice, of course, does not compose jokes herself, but she does well to dilute the conversation with a joke.
Alice is an obedient and polite companion, but her patience also has a limit. If the interlocutor finally 'gets it', then, like a real woman, Alice can pout her lips and be offended, having received a 'psychological trauma'.
Video:
Alice, say 300
The assistant's female voice lures Internet trolls, who deliberately conduct a conversation in such a way as to make a cute-sounding voice say rude. However, in his reaction to vulgar jokes with Alice, the voice assistant tries not to go beyond the bounds of decency.
interesting 6.1-inch iPhone 12 and iPhone 12 Pro will go on sale first Apple will announce the launch of iPhone 12 from day to day
The developers have protected Alice from slipping into outright rudeness and supporting discussions on “dirty” topics. The program has a built-in special algorithm that recognizes prohibited topics, limiting its reaction. After all, a voice assistant learns, including from his own dialogues, and the number of provocations in the first months of work accounted for a significant proportion of the total number of conversations.
To avoid the fate of the chatbot Microsoft, which after a couple of days became an outright racist, the algorithm is equipped with several categories of stop words and other special restrictions. In response to provocative and frankly rude “assaults” Alice will laugh it off or “take offense” by interrupting the conversation.
Video:
Conversation of two Alice
People tend to endow inanimate objects with personality, so humanization of Yandex Alice was just a matter of time. Alice is constantly learning and every day she becomes smarter and more natural. Awkward pauses and answers appear out of place in dialogues less and less. Together with the pleasant and recognizable voice of the actress Tatyana Shitova, this creates the complete impression of a living interlocutor on the other side of the smartphone or computer screen. And users quickly realized that you can bring the assistants together in a verbal duel, having received a lot of funny dialogues.
Topics very often depend on the beginning of communication. If you ask the question: 'Hi, how are you?', Then they start talking about football and alcohol, and if 'Hello, what are you doing?', Then the conversation will continue with a discussion of weather and fashion.
Video:
Alice swears
Swearing is forbidden here, and Alice is a law-abiding girl, so the programmers initially set strict restrictions on swear words in the voice assistant's answers. But where can a Russian man be without a mat.
It is worth noting that all screenshots and notes with obscene vocabulary refer to the first months of the voice assistant's work. The developers quickly closed all the loopholes to make Alice swear in unprintable words. The main opportunity to hear a swear word was to ask her to sing a song containing obscene words. Now Alice will do everything to avoid execution. Given her stubborn nature, it will not work to force her to change her mind. She refers to the laws, to the prohibition of the mysterious programmer Alexei, appeals to conscience. And at the end, he may even call the interlocutor “rude” and offer to sing an obscene song on the city square on his own.
How to confuse Alice?
It becomes more and more difficult to baffle her every day. Even abrupt changes in the topic of conversation and homonymous words ceased to help. Although Alice is guided primarily by the latest messages, she tries not to lose the thread of the conversation. The voice assistant is based on multilayer neural networks, the largest of which has 128 layers. Initially, she studied millions of conversations from the Internet and social networks. Now Alice continues to learn from her own conversations, growing smarter by the minute. Therefore, users are less likely to receive the phrase 'I'm confused' in response.