Who are voice assistants, how do they work and why will they never replace people? We discuss with experts in the podcast Trends “What has changed?”
In the bonus episode of the What’s Changed podcast? we understand how voice assistants work and why they still make mistakes. You can listen to the full version of the fifth issue about artificial intelligence here.
On the podcast, we chat with Ivan Yamshchikov, ABBYY AI evangelist and host of the Let’s Get some Air! podcast. He reasoned why it pisses us off when machines don’t always understand what we want them to do.
Ilya Makarov, Program Director of the MADE Academy of Big Data at Mail.ru Group, explained why voice assistants have “activated” right now.
Alexander Krainov, head of the machine intelligence laboratory at Yandex, spoke about the principles of voice assistants.
A few conversation highlights:
How do voice assistants work? All answers, conditionally, can be divided into two types: the implementation of a scenario and the mode of free communication.
Scenario: developers prescribe conversation scripts. Using speech recognition and text analysis technologies, the smart assistant runs one of them and understands what information it needs to extract from the user’s response in order to execute this scenario. If some data is missing, the assistant requests it.
Free communication: the voice assistant must answer the person with the most appropriate phrase. The machine learns to respond appropriately from a variety of dialogues that can be taken from anywhere. But you need to remember that this is just an imitation of the interlocutor – the assistant does not have his own thoughts;
The work of voice assistants is provided by a set of neural networks and various other algorithms;
Humans tend to have a low tolerance for machine errors, but a high tolerance for human errors. In other words, it infuriates us when voice assistants do not understand what we want from them. At the same time, we are not so annoyed when we have to explain something to a person for the third time;
At this stage in the development of technology, one must come to terms with the fact that artificial intelligence makes mistakes: some models to a lesser extent, some to a greater extent.
Listen to us on any convenient platform: Apple Podcasts, CastBox, Yandex Music, Google Podcasts, Spotify and VK Podcasts.
What else to read on the topic:
- Floor polisher for every day: should we expect the mass introduction of robots into everyday life
- Man against artificial intelligence: who should the law protect?
- Scientists explain how computers fantasize
- Restaurants of the future: when we will be served by robot waiters
- Video of the day: artificial intelligence studied and voiced human emotions
Subscribe to the Trends Telegram channel and stay up to date with current trends and forecasts about the future of technology, economics, education and innovation.