Replika, a chat app, was the focus of much of the content shared by members of a group called “Man-machine Love” on Douban. The app allows users to chat with and form close relationships with their robots. Quite a few sharers form a consensus that, in Replika at least, emotional interaction between a person and a program is possible.
But the group of people who hold similar views is quite small. “Man-machine love”, a very sci-fi color of the word, is still in films and literature and other works, far from our real life.
Returning to the everyday context, our relationship with the “machine” stays in the dimension of “dialogue” :
- “Hey siri, what’s the weather like tomorrow?”
- “Classmate Xiao Ai, how much is 123+321?”
We give instructions to the AI assistant, which takes them, parses them, executes them, and then reacts. Similar forms of communication are commonly seen in smart home and mobile phone applications.
With the development of ARTIFICIAL intelligence technology from time to time, especially the maturing of semantic recognition technology, the issues related to communication with AI have been frequently discussed. Typical among these voices is the concern:
When we get used to talking to AI in a commanding tone and getting a quick response, will this form of communication subtly influence our definition of “communication” and change the way people communicate with each other?
1. Why is our communication with AI directive?
In 2017, Apple released an AD for Siri that served as an example of a conversation between a human and an AI assistant:
“The Rock” Johnson plays the protagonist, through Siri to handle various things in life, including check the schedule, prompt items, call a car and other daily chores, he did not operate the phone screen, just issued such as “Hey, Siri, read my schedule! Such instructions.
Dwayne “The Rock” Johnson’s “Everything in Hand” commercial
Siri, the most iconic artificial intelligence assistant, was introduced in 2007 and bought by Apple for $200 million in 2010.
Siri as the representative of artificial intelligence assistant in smart home control, intelligent navigation and other areas of a wide range of applications, convenient daily life. People can control lighting by saying “Xiao Ai, turn on the living room light,” or use Google Home or Alexa to set up a kitchen shopping list and prepare dinner for the family.
But ai assistants are still relatively rudimentary. The beginning of technology investment is usually accompanied by admiration and doubt. For example, the aforementioned question: when AI is deeply involved in our lives, can we turn a blind eye to the imperative communication mode?
While “Hey Siri” seems to be able to handle a wide range of problems, a similar concern arises: could the command affect communication and, in particular, alter children’s speech behavior, creating an “ill-mannered generation”?
The question is, why do we use simple commands to communicate with AI? It has to do with efficiency, and it has to do with the innate purpose of science and technology.
In the human cognitive schema, the creation of science and technology is to expand human access to the outside world, bring benefits to human beings, and satisfy human needs from time to time.
As a technology genre, AI’s original intent and iterative approach is to make machines smarter and smarter to help users solve problems.
Just like we don’t ask it for advice before we use the broom. Because in our latent understanding, it is no emotion “tool”. Similarly, with an AI assistant, even though it can “talk” to us, we are clearly aware that these conversations are the result of an automated program. They’re still tools. The only difference is that they make some noise.
Therefore, unlike interpersonal communication, which often needs to interpret the hidden meaning of utterances, intelligent speech recognition technology is more about stopping responding to direct commands, that is, highlighting the “instrumentality”.
Wittgenstein’s theory of “Speech Game” holds that human speech, as a tool to transmit emotional information, takes symbols, words, patterns, music, pronunciation, body movements and facial expressions as carriers. The process of user and machine dialogue does not involve body movements, facial expressions and other “silent speech” to show the mood.
Therefore, when human beings invent code language and artificial intelligence stop interacting, even if smart speakers and mobile phone assistants can simply identify our mood stop, its essence is still tool use, or code operation, with program simplicity.
In other words, the process of communication between human beings and artificial intelligence does not produce emotional and other meaning sharing, and such verbal communication and dialogue cannot constitute a complete two-way communication and communication chain. The essence of user voice control is: to issue commands based on the purpose, which is no different from the previous command actions such as “click the screen” and “press the switch”, but voice control frees our hands and replaces them with verbal commands, which is the most efficient way at present.
If Hey,Siri becomes a role model for children
Synced has stopped summarizing ai failures in 2017, and its top 10 “failures” include:
When LG’s IoT AI assistant Cloi was first unveiled at the Consumer Electronics Show in Las Vegas, it failed to respond to the verbal commands of LG’s marketing director as an AI that could recognize the voice commands controlling household appliances.
Similar AI failures periodically remind us that so-called ARTIFICIAL intelligence may well be “artificial retardation”. Intelligent speech recognition technology, which relies on code operation, is far from replacing the complex and diverse theory of human speech communication, and may bring about many negative effects.
This is not groundless speculation or alarmist rhetoric. The basic idea of speechology is that context, meaning and environment will affect communication and linguistic habits. Benjamin Wolfe, an American speechologist, once proposed “Wolfe hypothesis”, which means that differences in speech structures will determine the perception and cognition of speech users of the world.
It is not hard to imagine that with more and more commanding words entering our daily life, people will gradually blur the boundaries between communication with machines and communication with people, thus bringing the habits of commanding words into interpersonal communication. As time goes by, maybe in the future, common words such as “please” and “thank you” will be on the edge of extinction.
More importantly, the problem may have a more pronounced impact on growing children’s verbal communication. Psychological research generally think: children since birth that is to see the world, in a long period of time through the simulation of others’ behavior (expression, speech, behavior, thought) to obtain growth, their differentiation ability is relatively weak, unable to make discrimination, selection and elimination, it is likely to be completely learned.
Bandura, a famous American psychologist, once conducted an experiment called “Bobo doll”, in which children learned violent behaviors by simulating the violence of “paragons” on the doll. Therefore, he thought that children’s social behavior acquisition was mainly completed by observing and simulating the behaviors of important figures in their ideal life.
Bandurabobo doll experiment
This is borne out by a guangming.com news report, in which a parent reported that after watching the animated film Boonie bears, his child was learning the characters’ words and cursing words such as “stinky bear”. Media outlets have pointed out that the hit CARTOON Boonie Bears on China Central Television once used 20 inculturalwords in more than 10 minutes of content. It is reported that under the influence of children’s films and TV series, there have been cases of children imitating the violent situations in many places. As a result, cartoons such as Boonie bears were later made suitable for 10-14 year-olds.
Similarly, the original intention of the FILM rating system in the United States was to be formed by a committee of parents, which stopped rating movies based on themes, language, violence level, etc., to help more parents determine which movies are suitable for children of a certain age.
Compared with concerns about the impact of violent language in movies and television on children, the possible negative impact of communication with AI on verbal communication seems to be far from arousing the vigilance of parents, and the academic community has not yet reached a mature conclusion.
But these negative effects cannot be ignored. When children stop communicating with artificial intelligence for a long time, under the manufacture of “simulated” learning machine, it is inevitable that changes will occur in the use of speech, and it is getting closer and closer to mechanized communication like AI.
Directive speech violates the “politeness criterion” in verbal communication, that is, the speaker should always try to give others as much as possible when speaking, so as to make both parties feel respected in communication, and in turn get the other party’s good impression on him.
On the contrary, the power difference and impolite behavior in the communication mode of command discourse will cause communication failure to a large extent, and even make the communication object stay away from you.
Imagine a child talking to parents and friends with phrases like “Get me the toy” and “help me turn on the game”; Say “Hey, take me to the center” when a stranger asks for directions. To ask a colleague for help and say, “Help me with this,” such a command tone, at least for now, no one but AI seems to pay attention.
3. How is dialogue with AI possible: The expansion of communication borders and meaning sharing
Of course, it is not unknown that the above can lead to negative changes in verbal communication.
Many technology companies at home and abroad have stopped optimizing related products for fear that the way of communicating with artificial intelligence will be used by users, especially children, who will become “impolite”.
In 2018, for example, Amazon and Google each took part in encouraging polite phrases in their smart assistants. For example, when people use please, the assistants respond with “Thanks for asking so” as a compliment.
In terms of achieving better emotional interaction between AI and humans, technology represented by AI mood research is trying to make machine-human dialogue more intelligent and deal with more complex human-computer interaction.
The pause in this category is mainly manifested in two aspects: one is to enhance the AI’s perception of the underlying mood patterns of human beings by recognizing the speaker’s words and expressions; The second is to enhance the AI’s responsiveness to human emotions.
Similar improvements could, of course, make the machine less of a “tool man” and more sympathetic to its users in the process of communication. For example, the fifth generation of “xiaoice” developed by Microsoft is different from previous generations. Its function is not limited to simple chat reply, but also can have independent speech organization ability. Microsoft even let it break through the constraints of virtual technology to send users Christmas wishes by phone. Xiaomi’s Version 5.0 has also become more intelligent, proactive and humanistic.
As a result of these efforts, ARTIFICIAL intelligence seems to take on emotion, allowing users to feel its emotional feedback rather than seeing it as a cold-spoken machine, effectively preventing children from falling into the simulation and habit of commanding conversation.
The problem is that simple emotional recognition and response through a machine is not enough to deal with the transition from command to speech.
In the fundamental concept of communication science, human “communication” is endowed with the meaning of “sharing meaning”, and only interpersonal communication can produce emotion and meaning exchange.
In the face of a cold machine, it is difficult for us to attach emotions such as “politeness” and “love” to verbal communication. Therefore, if you want to make the human-computer interaction more close to the interpersonal communication exchange and sharing, should first face differences that exist in the exchanges between the two, and then from the technical level to enhance the artificial intelligence of dialogue of information disposal and identify ability, makes the prediction of AI from time to time to stop in the interaction, adhere to and guide, at the same time provide human possibility and way to understand the algorithm of communication.
As AI technology evolves from time to time, the way of speech communication between human and machine becomes more and more diversified. In the face of new technology and its associated communication problems, the meaning of “communication” will be redefined and the direction of “meaning sharing” will be expanded. A more ideal impact might be that communicating with AI will force us to rethink the boundaries of communication and the social meaning of verbal communication.