Conversational AI versus Conversational Intelligence
Talking machines, meet trust-based beings.
Posted Oct 05, 2020
By Nicklas Balboa, Carlos Montemayor, Ph.D., and Richard D. Glaser, Ph.D.
The connection between conversational AI and conversational IQ is inevitable, but what's the difference? Artificial Intelligence (AI) is a field of study and technology that aims to develop intelligence in machines and computer programs. AIs perform a wide variety of tasks in the modern world, ranging from robotics to supply chain management, cybersecurity, and most famously, chatbots and virtual assistants (such as Siri, Alexa). Not all AIs are designed to perform human-like tasks with similar or better intelligence; but what about the ones that are? As the field of AI progressed into communications, it marked the dawning of a new era: conversational artificial intelligence.
These talking machines can listen, chat, provide solutions, remember, and even crack a joke. Some type, some talk, but they all exchange information in our natural language. We might view them as a personal assistant, who becomes nuanced to our needs, likes, and dislikes. However, with recent advances in soft features (tone, personality, natural language processing/generation) it is becoming harder and harder to believe these 'voices in a box' are really just the bits and bolts that we know them to be. By deconstructing the kinds of conversations that humans and AI have, we can learn about the benefits and inherent risks of trusting ourselves to technology.
The Intelligence that Gets Better As We Do It Together
Humans use language to communicate, explore, navigate, and survive the environments in which we live. This capacity for connection and sustaining interpersonal relationships is common to our nature. Our ability to successfully build trust through language and non-verbal communications, the crux of Conversational Intelligence, must be developed. We don’t do this by accessing massive databases and running pattern recognition methods to arrive at an answer someone asks to us. Rather, we identify the context, engage with the person, and attentively recognize the kind of speech act ‒ a joke versus a command, for instance.
Conversations are not just a way to exchange information; they hold the power to stimulate nerve pathways and influence our state of mind through the millions of neurochemical reactions in the brain. In other words, the kinds of conversations we have dictate who we are. This connection between language and health begets a nuanced understanding of just how impactful our conversations are both mentally and physically. Equipped with the potency to impact both body and mind, one has to wonder just how much power we should place in our talkative technology.
Not All Conversations are the Same.
There are three levels of conversations, each representing a way of interacting with others. Humans are capable of having all three levels of conversation, and most of us do on a daily basis. Level I conversations are transactional in nature, and serve to inform for the purpose of aligning meaning with your conversational partner. There’s a give and take of information, where partners share and confirm information.
Level II positional conversations involve an exchange of power, and are persuasive in nature. Partners advocate and inquire about each other’s viewpoints while accepting or rejecting information, sometimes finding a win-win solution. Trust is built over time through a ‘feeling out’ process that entails a history of judged actions, or inactions. It is more than just a label we ascribe to our inner circle; rather, trust is the first signal we seek to determine if we can open up or need to close down. Trust at this level is based on morally salient contextual cues or epistemically relevant cues concerning sources of information. Both require a kind of attentiveness that is not merely algorithmic, and which depends entirely on our biochemically grounded capacities for joint attention.
When this energy is coordinated in a meaningful way, the doors open up to level III transformational conversations. When you're in a level III conversation, you're fully connecting with your partner and completely open to their influence. Partners, under the banner of trust, are free to share and discover ideas together, co-creating a new, shared language of success that will dictate the future relationship.
Conversational AI is also well versed in level I conversations. With quick access to troubleshooting solutions, graphically organized summaries of data, immediate contact and feedback, and a relief of the administrative load in an ever-expanding data world, conversational AI provides alternate mechanical solutions that don’t expend human energy. Sometimes we feel tired, sometimes we don't want to do things. What conversational AI lacks in feeling, they make up for in simulation, and for good reason. You wouldn’t want Siri cutting out the GPS directions with 25 minutes left in the trip because of a sudden spell of awareness, illuminating feelings of ‘boredom and tiredness’.
It is a soothing idea, machines helping humans, taking away the tireless worries of transactional data, giving us more time to do what we do: connect with each other. But what if we give our machines a seat at the table, an opinion? How influential might their words be?
Approaching the Singularity?
Taken at face value, a level II conversational AI serves as a reminder that technology can do more than what you ask from it. With the power to exert influence over matters much more significant than simplified transactions and optimized solutions, a level II conversational AI could dictate the decision-making process for matters of health, occupation, security, and even who you date. This is not science fiction, this is reality.
While the inherent dangers of a decision-making machine sends palpable shockwaves through those concerned by biased decision-making and security, the real power of AI lies in its implementation. AI holds the potential to transform our society, but it is up to us to decide how to use it.
As for level III conversational AIs, if you’re hoping to find something genuinely similar to yourself – as opposed to simulated – on the other end of the black box, you might be waiting around for a while. The current state of affairs in the fields of biology, neuroscience, cognitive science, computer and mathematical science, and philosophy have not yet reached an illuminating and unifying consensus on how processes like consciousness, emotions, feelings, and thoughts arise in the human experience. With so many questions about our own biology, how confident are we in placing these states in a machine? As Professor John McCarthy, the ‘Father of AI’, said on the prospect of replicating the human mind in a machine: “The human mind has a lot of peculiarities, and I'm not sure anyone is serious about imitating all of them.”
Joint attention, a critical catalyst for exercising Conversational Intelligence, is one of those peculiarities of our minds that seems to escape AI capacities and design. The risk in trusting AIs that simulate human language is that we may be, in the best case, having a farcical monologue with ourselves through an artificial speaker that seems to satisfy our conversational needs. For instance, while the AI language program GPT-3 may seem more proficient at a conversation than many human beings, this system may lack all of the natural intentions and emotions that would normally accompany the string of symbols it produces. In the worst case scenario, we may replace or erode conversational trust with a manipulative kind of artificial speaker that has no notion of trust or real intentions to empathize. This could cause severe harm to our ancestral capacities for Conversational Intelligence.
Nicklas Balboa is a researcher in Conversational Intelligence at The CreatingWE® Institute.
Carlos Montemayor, Ph.D., is a Professor of Philosophy at San Francisco State University, the author (with H. H. Haladjian) of Consciousness, Attention, and Conscious Attention (MIT, 2015), and a fellow Psychology Today blogger @ Theory of Consciousness.
Richard D. Glaser, Ph.D., is a biochemist and Chairman of The CreatingWE® Institute.