The journey to human-machine symbiosis

What is human-machine understanding and how can it improve human-AI interaction?

Current technology is great at understanding a task, but struggles with understanding the human behind the task. Through HMU development, we can create a new wave of technology that can master both, revolutionising how we live, work and achieve our ambitions.

To do this, a deep tech interdisciplinary approach is essential. CC’s HMU team has brought together psychologists, cognitive scientists and AI experts to create technology that can interpret human behaviour; what led to it and what comes next. Our HMU models process human behavioural data, such as motion, kinetics, eye movements and pupil dilation to infer what's driving human behaviours to build the future of human-machine interaction.

This work ultimately aims to deliver personalised machines that get to know you and adapt to you over time, which could be game-changing across a huge range of use cases.

Take digital surgery as an example. Digital surgical platforms already integrate robotics, surgeon interfaces and data to help perform very precise surgical procedures – but as powerful and robust as these tools are, they are operated by humans. Humans have skills these surgical tools don’t, but we also have many variabilities, resulting in unpredictable outcomes. After all, even top surgeons can have off days, whether due to stress, fatigue or otherwise.

By incorporating HMU into this technology, we can take the precision of digital surgery a step further by factoring in the state of the human surgeon at any given time, utilising the best of both human and machine skillsets. By interpreting data to assess a surgeons' mental and physical states, the machine can ensure they are comfortable, focused and effectively collaborating with their team to minimise risk and elevate the chances of success.

When done correctly, this allows for us and machines to work intuitively together in a way we never have before. Machines can become an extension of human teams, elevating their unique skillsets while accounting for their intrinsic humanity.

The challenges of developing empathetic, interactive AI

To make this concept reality, we must first understand the challenges. The first and potentially biggest challenge is an obvious one: how to build empathy into AI. Empathy is a complex but integral part of humanity, allowing us to adapt to each other’s needs and gain a shared understanding of how someone feels in any given situation. This shared mental model is essential for us to interact effectively.

Machines aren’t currently designed to do this. At best, they react to specific aspects of our behaviour, but don’t try to infer what these behaviours might mean or what led to them in the first place. Due to this gap, the machines we’re used to working with have a very limited contextual view of how we’re feeling and how it might affect our behaviour. This divide is what we are aiming to bridge through human-machine understanding.

Empathy and understanding human needs are only one side of the coin - the next challenge is identifying how to respond based on that understanding. There are already promising results in specific, limited scopes, but developing a general model is challenging – in part because understanding human behaviour and establishing empathy can only be done in context.  Humans can naturally understand the meaning behind each other’s actions through familiarity with human behaviour and knowledge of different contexts – but teaching technology to do the same will take time and research.  

Here, we can rely on the myriad of human-machine interfaces already available and apply our expertise to identify how best to support in the right way. In our digital surgery example, this may mean the system adjusting haptic feedback, modifying audible alarms or adapting visuals cues based on the surgeon’s needs.

It is also crucial that ethics and trust are built into these empathetic machines. Deeper machine understanding of humans relies on deeper communication and observation – something that cannot be achieved without raising privacy and ethical concerns as machines gain more agency, potentially making decisions for us or influencing our behaviour.

CC is tackling these issues head on through our deep tech approach to human-centric AI assurance, founded on pillars of safety, ethics and security to ensure measures are integrated from the beginning of any AI development for maximum security. Through this, we can maintain a balance between challenges around safety and privacy with the positive impact HMU can have on business, society and the planet.

The commercial promise of human-machine understanding

This may sound like science fiction, but the reality is closer than you think. And as well as having a massive impact on the relationship between humans and machines, HMU also presents a huge commercial opportunity across a range of sectors – from healthcare to consumer and industrial.

At this pivotal moment, businesses have an exciting opportunity to craft strategies that will maximise human potential by seamlessly integrating their human teams with technology. While the future may be driven by artificial intelligence, it must be designed with human intelligence at its core.

Ali Shafti is the head of human-machine understanding at Cambridge Consultants. He holds a PhD in robotics with a focus on human-robot interaction and has more than 10 years of experience in research and development for human-machine interaction.