“Are you human?” This is the question that will soon follow “Good morning how are you ?“, according to Laurence Devillers, professor of artificial intelligence (AI) at the Sorbonne University and researcher at CNRS, specialist in human-machine affective interactions, in a context in which exchanges between brands and consumers are increasingly automated. Why the conversational bots they are now an integral part of the relationship with the customer, both oral and written: very often first of all, to provide a first level of information at any time, or on any day of the week, and thus contribute to greater availability of the brand. However, as the line between human and machine interactions becomes blurred, the sustainability of a healthy brand-customer relationship can be called into question.
Robots that adapt according to their interlocutor
“The trend is towards finer sensing of all implicit and emotion by conversational agents, particularly during a call, in order to better personalize interactions and adapt to the customer’s” mood. “AI is therefore more. humanized, in the sense that it gives the impression of understanding and resonating with what the client is expressing ‘, explains Laurence Dellers, met during One to One Biarritz. Quotes from artificial intelligence experts 3 essential elements for a conversational robot to empathize : the presence of sensors to detect emotions, that of processors to analyze the parameters collected and make a decision and, finally, “actuators” to act with the real world and express an emotional response.
“However, remembers Laurence Dellers, artificial intelligence has no conscience, no morals. This is a fact to always keep in mind! ” It is therefore better not to get too attached or to attribute no faculty of feeling to the AI. A “drift” that we find, however, in Japan in particular, with the initiative Gate who offers to purchase a smart assistant presented in the form of a hologram enclosed in a box and which can quickly be considered a “virtual girlfriend” after a few weeks of use.
Nudge marketing: opportunity or risk?
“Since artificial intelligence continually adapts to its interlocutors, the challenge now is to make sure that it is truly a consideration of a context, in an ethical way, and not an incentive to behave in particular, such as the move to purchase. “, continues the researcher. The teams responsible for developing and monitoring conversational agents must therefore always be vigilant with respect to the answers provided to customers, to ensure good relevance on the one hand, but also the absence of a certain incipient manipulation towards a type of reaction or behavior.
The importance of standardization of practices
Risk of “black box” effect, possible prejudices linked to non-equality between men and women – 90% of programmers are male while 80% of RNs have a female name – or poor quality of the data presented for learning … “There are many flaws related to the use of AI in customer relations, hence the importance of setting rules that apply to everyone. One of the proposals in particular is to implement random gender naming when we create a conversational agent, which therefore can be both male and female depending on the draw, but in reality this restores a certain correctness “, suggests Laurence Deviller. And to conclude: “In a future world in which we will be increasingly inclined to dialogue with artificial intelligences, the global challenge is precisely that of a public awareness, adults and children alike, about what artificial intelligence is. We must constantly remember that even if the machine seems more and more embodied, it still remains devoid of conscience and feeling.
