AI for emotion detection: an industrial challenge
June 10, 2022
The presence ofvoice assistants in consumers' interactions with customer services is intensifying. The latter are responding in an increasingly precise and personalized way, thanks to the lightning progress of artificial intelligence (AI), in terms of speech recognition. What's more, the detection of emotions in spoken language (emotional AI) represents a major challenge. The better a voice assistant understands the emotional state of its interlocutor, the better it can adapt its interaction.
Faced with the sheer volume of inquiries, and the systematic need to use a customer interface to find answers, companies have understood the stakes and equipped themselves accordingly. AI is invaluable when it comes to dealing with an "industrial" influx of questions, qualifying requests, directing customers and thus guaranteeing flawless availability! And we see it every day: our favorite digital tools are fed with AI, and this contributes greatly to the trust we place in them. So, where do we stand?
What is emotional intelligence?
Emotional intelligence is based on "affective computing". affective computing "which consists in segmenting human emotions. Several theories exist on the subject. One of the most widely used is the Plutchik wheel, which comprises eight basic emotions.
They are structured in opposing pairs:
Joy and sadness,
Fear and anger,
Disgust and confidence,
Surprise and anticipation.
These basic emotions are now integrated into emotional intelligence engines, enabling voice assistants to conduct realistic discussions, and voice assistants to have a form of sociability, even if it remains elementary! We haven't yet reached the stage of "thinking machines"... "sophisticated reasoning" is the next challenge for artificial intelligence! Nevertheless, voice assistants are capable of detecting some of our emotions, interpreting them and even reacting to them!
Artificial intelligence detects our emotions
To achieve this, artificial intelligence relies on various types of sensors: video (recognition of emotion read on the face), microphones (analysis of voice tonality) or biometric sensors (e.g. wristbands analyzing nerve pulses). The customer's level of stress, anger or satisfaction is thus assessed in real time, and the voice assistant's reaction parameterized accordingly.
Artificial intelligence understands and interprets our emotions
Once an emotion has been detected, artificial intelligence analyzes it, taking into account context, language, cultural environment and, when perceived, voice and intonation (for sound sensors) and gestures (for video sensors). These machine-learning-based techniques aim to adjust the nature of the voice assistant's response precisely according to the emotional context in which the dialogue takes place...
Artificial intelligence acts on our emotions
In addition to adapting its responses, the voice assistant can also adapt its "behavior". If the conversational assistant has detected that the user is in a hurry, it can speed up its speech rate, and generate more succinct answers. The voice agent can also accumulate memories of its interactions with the user, so that it doesn't have to re-explain its case with each new interaction. Finally, if it feels overwhelmed by the complexity of the question being asked, or by a state of anger or stress that is too intense, it can also decide to refer the call to a human agent, ultimately called an augmented agentwho - at the same time as the call - recovers all the contextual scenario and data.
So why not reconcile these two intelligences?
In the future, a customer service department will undoubtedly need to combine the strengths of AI with the subtle gentleness of emotional intelligence. At once understanding and dampening customer emotions, and delivering a hyper-relevant response in record time, anywhere and anytime. Of course, this raises a question of ethics... Will the use of emotional understanding in customer relations be to better respond to needs or to try to influence decisions? Today's customers need to express themselves, ask questions and be understood. They like to learn by getting answers, and to be given special attention. Constant progress is being made to use artificial intelligence technologies wisely and emotionally to deliver excellent service.
At ViaDialog, we are constantly striving to offer our customers state-of-the-art technological solutions for maximum customer satisfaction. In particular, this means supporting agents in their interpersonal relations, for example, through indicators of emotional overload. To achieve this, we maintain an ongoing scientific watch to ensure the emergence of ViaDialog products at the cutting edge of technological innovation!
Would you like to find out more? Our experts are available to answer your questions!
