Photo by Mulyadi on Unsplash
In the rapidly advancing world of artificial intelligence, one challenge stands tall: the quest for machines to truly understand and express human emotions, and in so doing both read and create personalities – to move from the mimic to the relatable. This challenge – the current lack in depth of interaction between humans and machines – signifies the gap between the currently limited (if impressive) AI, and the ultimate goal of achieving Artificial General Intelligence (AGI).
Emotion: More Than Just Data
One of the primary hurdles for AI in understanding emotions is the intrinsically nuanced nature of emotions themselves. While AI excels at processing vast amounts of data and recognising patterns, emotions are not just data points. They are complex, multidimensional, and deeply intertwined with human experiences, memories, and culture. They often appear fickle on the surface, but really deep underling beliefs.
Unlike a mathematical problem, there’s no definitive solution to understanding or expressing an emotion. It’s an increasingly accepted preemies that the human brain works as a predictive tool, but importantly a tool with the ability to rapidly create new neural pathways when reality doesn’t meet our predictions. We have the ability not just to predict, but to continually update our model.
The prevailing neural network models strive to replicate this prediction approach, but, they overlook one key element: the unique influence of mirror neurones in a humans brain. It’s mirror neurones that enable us to recognise, and empathise with, a reaction observed in others, by drawing from our past experiences. This is what makes us human and allows us to form relationship. It build our personalities and is a evolutionary output of our social natures. And importantly it’s far, far more nuanced than current models can replicate, however large they grow.
Real-time Interaction: The Empathy Gap
When it comes to real-time human interactions, the absence of genuine empathy in AI becomes glaringly evident.
Imagine a doctor, delivering a life altering diagnosis to a patient: In such a deeply human moment the doctor must strike the balance between clarity and comfort, between closeness and professionalism. Endless nuances, built up through their unique interpersonal relationship, must be subconsciously analysed to personalise this, and to adjust on the fly as their patient reacts. This is not a ‘generate and deliver’ but iterate and adapt.
Setting aside the mode of communication momentarily (e.g. the absence of a human visage), a chatbot could perhaps deliver a ‘technically correct’ response, and signpost support, but the palpable deficit of genuine emotional depth would render the exchange hollow and detached. This dearth of adaptive empathy, and the nuanced calibration of messages, curtails AI’s efficacy in any role demanding emotional acumen; from mundane interactions to sculpting a truly resonant marketing campaign.
Bridging the Gap: The Path to AGI
AI’s ability to discern and express emotions will be a significant milestone on the path to AGI. While current AI systems can emulate, and perhaps out perform in, tasks that require intelligence, AGI will be expected to perform intellectual tasks that humans can and, crucially, to be capable of relaying its responses in a manner that displays understanding of the human receivers emotions. Cracking the challenge of emotions is therefor not just about improving human-AI interactions—it’s a fundamental step in achieving AGI.
The Future: Hope and Challenges
While the potential is immense, the journey is fraught with challenges. Teaching AI to understand emotions requires more than just algorithms and data. Indeed it can’t be about teaching and has to be about iterative learning, and to facilitate this the ability to feel emotions too – senses of delight and shame fundamentally shape our inner understanding of self. Mirroring this demands creating an intuitive deep understanding of human psychology, culture, and individual experiences. Ethical concerns around AI are a hot topic, but we’re only at the tip of the iceberg: when machines learn to respond to, and perhaps to manipulate, human emotions, the challenges become immense. We will move past discussing regulation into re-defining morality.
Is a machines ability to understand emotion blocked only by the lack of appropriate data ‘inputs’ to our models? Or is there an ethereal quality to personality that’s uniquely human? Will we really be able to distil the essence of our ‘sole’ into a statistical model? And should we really want to?
AI has made significant strides, but the realm of emotions remains a formidable frontier – and I’m not sure it’s one we want to fully cross. As we continue to explore this territory, the hope is for a future where machines not only understand our world but one where they understand the intricate tapestry of emotions that define the human experience. Perhaps true emotional intelligence, not just mimicry and manipulation, is the bridge that will keep machines as augmentations to our lives, and not replacements for true human interaction.