I’ve always been a voracious reader, moving from classic novels to the latest cookbooks without missing a beat. Earlier this year I merged that habit with my fascination for technology by building a tailor-made GPT I called “Find Your Next Great Read.” The bot asks a few questions about genre preferences, looks for recurring themes, and then returns a short list of titles it thinks could spark the next literary adventure. I shared the project with a friend, expecting a quick chat about hidden gems, but the conversation pivoted within sixty seconds to a 60 Minutes segment on humanoid robots and the looming dangers of artificial intelligence. The abrupt shift made me realize how swiftly the topic of AI can dominate any discussion.
This detour sparked a deeper reflection on the way we develop opinions about a technology that feels both futuristic and unfamiliar. When fear dominates, we tend to notice stories about cheating, job loss, privacy erosion or the spread of misinformation. Conversely, optimism draws us to headlines about medical breakthroughs, productivity boosts, or creative collaborations. Both sides can be built on solid reporting, yet each can obscure the broader landscape. Such selective attention is a classic case of confirmation bias, a mental shortcut that nudges us to give extra weight to information that already aligns with our existing beliefs. Recognising that tendency is the first step toward a balanced viewpoint.
How bias shapes our perception of artificial intelligence
Our brains are wired to protect the ideas we already hold, so when we hear a story that confirms a pre-existing fear or hope, it sticks like glue. The confirmation bias works both ways: alarmist reports about autonomous weapons amplify anxiety, while glowing accounts of AI-generated art fuel excitement. Because the media often highlight the most dramatic angles, the average person ends up with a fragmented picture—either a dystopian nightmare or an unstoppable utopia. Recognising the swing of this bias encourages us to actively seek out counter-examples, such as case studies where AI improved safety in hospitals or where safeguards prevented misuse.
What I learned by testing AI tools first-hand
My first encounter with ChatGPT a few years ago left me underwhelmed; the generated blog drafts were generic, and the promised month’s worth of social-media copy felt forced. I set the tool aside, only to return months later when a work project hinted at a useful application. This second round revealed a surprising shift: the model now suggested nuanced angles, spotted patterns in my own arguments, and even challenged assumptions I hadn’t considered. The most valuable lesson was that AI can act as a mirror, reflecting hidden connections in our thinking. Yet the mirror sometimes distorts—confident instructions have failed me when the underlying data were outdated, reminding me that a persuasive tone does not guarantee accuracy.
Three low-risk ways to explore AI yourself
Turn a hobby into a test case
Pick an activity you already enjoy—gardening, cooking, or reading—and ask an AI model for tailored advice. For example, describe the shade conditions of your backyard and request plant suggestions, or list a few beloved novels and ask for off-beat recommendations. Compare the output with trusted sources or personal experience. This simple experiment shows how well the system captures context and where it invents details, letting you gauge reliability without exposing personal data.
Let the AI decode a manual
Download the PDF of a household appliance you own, such as an air-fryer or a coffee maker. Upload the file to a conversational AI and request a plain-language summary of a specific feature. Then read the official manual section yourself and note any discrepancies. This exercise highlights the model’s ability to simplify technical jargon while also exposing its tendency to hallucinate facts when the source material is ambiguous.
Use AI as a decision-making partner
When a routine choice feels burdensome—like planning a holiday gathering or reorganising a weekly schedule—share the relevant constraints with the model and ask for a handful of alternatives. You may discard most suggestions, but a single unexpected angle can unlock a solution that resonates with you. The key is to treat the AI as a brainstorming assistant, not a final arbiter, and to verify any actionable advice through a reliable channel.
These low-stakes experiments illustrate that curiosity, combined with a pinch of critical thinking, lets anyone build a personal sense of what AI can and cannot do. By testing the technology directly, questioning its answers, and cross-checking against trusted references, we can move beyond headline hysteria and make informed choices about the role AI should play in our daily lives.
