In a galaxy far, far away… few partnerships are as iconic and enduring as that between Luke Skywalker and R2-D2 from the Star Wars saga. Their dynamic relationship embodies the essence of collaboration, trust, and mutual reliance and is the inspiration and aspiration behind my passion for human-machine teaming research.
While achieving a dynamic relationship akin to that of Luke Skywalker and R2-D2 may still be a prospect that is far, far away, we are steadily progressing towards more sophisticated forms of collaboration between humans and machines.
But we are not there yet.
The dynamics of teams of people and human-machine teams have discernible characteristics that result in observable differences in how they collaborate.
A designer must be cognisant of these distinctions when designing for human-machine teaming scenarios.
The following observations about team characteristics and possession are based on my research on human-machine teaming:
Understanding the Dynamics of Human-Machine Collaboration
Both teams of people and teams of people and machines engage in verbal exchanges. With the advent of conversational user interfaces, machines can now engage in natural language conversations, mimicking human-like communication patterns and understanding context to a certain extent. This development has led to a convergence of communication styles between humans and machines.
However, while human-machine communication typically involves specific commands or queries, human-human communication incorporates a wider range of verbal expressions. Additionally, non-verbal cues like body language and facial expressions, currently not used in human-machine teams, play a significant role in conveying emotions, meaning and intentions, enhancing interpersonal understanding and connection.
In terms of contextual understanding, people possess the ability to grasp subtle contextual cues, body language, sarcasm, irony and the meaning behind pauses and silences, allowing us to interpret and respond to other human beings. In contrast, machines are limited by their programming, architecture, training and available data.
Teams of people often make decisions collaboratively through discussions, arguments, opposition and consensus, drawing on critical thinking, intuition and experience. In contrast, machines follow programmed decision-making processes, excelling in making data-driven decisions quickly and consistently but lacking in autonomy and accountability.
Interpersonal trust, the foundation of human interaction, is built through social interactions, mutual understanding, reflection and emotional bonds. While humans may trust machines based on their consistent performance and reliability, this form of trust lacks the emotional depth, reciprocity and vulnerability characteristic of human interactions.
Furthermore, the dynamics of conflict resolution differ in human-human and human-machine teams. People resolve conflicts through explanation, negotiation, persuasion and compromise, leveraging interpersonal skills such as empathy and emotional intelligence. In contrast, machines do not possess these capabilities, relying solely on programmed algorithms to address conflicts or discrepancies.
How Might We Improve Human-Machine Collaboration?
The evolution of AI is progressing rapidly. These advancements aren’t just technological; they encompass ethical considerations, interdisciplinary collaborations, and the integration of AI into domains that were previously resistant.
Looking forward, I’m eagerly anticipating significant advancements in natural language processing coupled with emotion recognition capabilities. I’m expecting this to lead to a fascinating convergence of human-human and human-machine team dynamics. For example, in my work with Easol Ltd, we are developing a large language model trained on mission-relevant information coupled with a conversational user interface that has the potential to adapt its multimodal outputs upon recognising emotional tones in people’s voices. This adaptive interface aims to improve interactions and provide more empathetic and contextually appropriate responses.
Furthermore, there is immense potential for machines to learn from domain experts, who possess invaluable insights that cannot be found in textbooks or databases. Who better to impart knowledge about a specific field than the individuals who have devoted their lives to it? At Tulpa Ltd, I am engaged in combining the power of AI with human wisdom, a process involving meaningful dialogues with experts and structuring the insights into machine readable format. Leveraging this wisdom, we iteratively refine AI models trained on the expert-acquired data. Throughout this process, experts validate and fine-tune the outputs of AI systems, ensuring accuracy and relevance.
Our vision is a future where AI systems can actively seek explanations for human behaviour, fostering mutual learning and collaboration and enabling machines to learn from people’s actions. AI systems that question data, solicit multiple perspectives, and address biases in real-time could lead to decisions that are more equitable and just. Tulpa Ltd aims to achieve this with their human-machine teaming platform.
While we may not yet have fully realised the dynamic relationship of Luke and R2-D2, the companies I’m working with are continuing to shape AI that might some day reflect our intelligence and uplift our humanity. Please get in touch to find out more!