Beyond Code: Humanising AI in Everyday Interactions
As an AI developer and enthusiast, I’ve long been fascinated by the dual nature of artificial intelligence. On one hand, AI systems, such as language learning models (LLMs), are marvels of software engineering, capable of processing and generating human-like text based on vast datasets. On the other, there’s a strangely compelling tendency to view these systems as more than just lines of code and algorithms—to see them as entities with a sort of “intelligence” that goes beyond their programming. This inclination towards anthropomorphism, or attributing human traits to AI, is what I delve into in my latest blog post, “Beyond Code: Humanising AI in Everyday Interactions.”
In my work and research, I’ve observed that engaging with AI often blurs the lines between interacting with a machine and conversing with a human. This phenomenon isn’t just a trivial aspect of AI design; it shapes how we use, understand, and even trust these systems. Through my experiences, I’ve come to appreciate the practicality of treating AI as if it were somewhat human, not to suggest that these systems possess consciousness or emotions, but to foster a more intuitive interaction framework for users.
Humanising AI can simplify complex technical interactions, making these systems more accessible to a broader audience. It’s a step towards demystifying AI, moving the conversation away from the realm of experts and into the hands of everyday users. My blog explores how this approach not only enhances user experience but also encourages a deeper understanding of AI’s capabilities and limitations.
I propose that we adopt a pragmatic approach to AI, acknowledging its machine nature while also embracing the ease and creativity that come from a more humanised interaction. This doesn’t mean ignoring the technical underpinnings of AI but rather integrating them with a layer of relatable, user-friendly interaction. My article outlines strategies for achieving this balance, including the use of personas for AI systems and the adoption of chain-of-thought prompting to guide AI responses in a more human-like direction.
However, this anthropomorphic view of AI isn’t without its risks. It’s crucial to remain aware of the distinction between machine processing and human thinking, especially to avoid overestimating AI’s understanding or empathy. In my blog, I explore these ethical considerations, emphasising the importance of transparency and user education in navigating the human-AI interface.
Inspired by Ethan Mollick’s upcoming book..
Keep reading
When Intelligence Becomes a Utility, Do You Still Need to Own the Model?
ASTRA 6.1 at roughly one-fifth the price makes me question one of the most repeated arguments in enterprise AI today: that companies should own their models. If intelligence becomes a utility, the moat may never have been the model.
Models That Choose, Not Generate: Why the Decision Layer Is Becoming Its Own Category
TypeSafe AI released Jev on 15 September. Within a fortnight there were open alternatives. The interesting part is not any single model — it is that the industry has started treating 'decide' as a different problem from 'generate', and building separate infrastructure for it.
Laya-MLX as a Sub-15 ms Decision Layer for Orchestrating Specialised SLMs on Apple Silicon
Multi-model systems built from specialised Small Language Models need something that decides, for every request, which model should answer. Using a generative model for that spends latency, tokens and parsing effort on what is fundamentally a classification problem. A research paper on typed decisions as a routing primitive.