I’ve instructed thousands on utilizing AI – here’s what I’ve discovered.

In my journey of training teams to adopt AI in their workflows, I’ve taken notice of a burgeoning professional divide that is evolving uniquely. It has become apparent that while some individuals fully surrender their tasks to AI without a second thought, others completely reject its use for fear of its limitations. However, within these extremes lies a third group. This group understands how to collaborate with AI critically, treating it like an eager intern that requires oversight, encouragement, and careful management to excel in its functions.

The major differentiator among these groups tends not to be technical skills but rather an attribute that seems increasingly rare in our fast-paced world: curiosity. Those who exhibit a genuine willingness to explore, to learn from mistakes, and to discern the true capabilities of AI are thriving. Here is what I have gleaned through my experiences so far.

Most individuals struggle with AI because they lose sight of its true essence

My time working with teams has revealed that people often oscillate between extremes when engaging with AI. They either treat it as an omnipotent oracle or dismiss it entirely after encountering just one hiccup. The reality is that today’s AI technology shares as much commonality with the human brain as a bird does with a modern aircraft. While both may facilitate ‘flight’, that’s where the similarities begin and end. Large Language Models operate by predicting words based solely on patterns they observe in their training data. This explains their ability to generate coherent descriptions on common subjects, while also leading them to confidently fabricate information when navigating less familiar contexts.

When users grasp this fundamental truth, the way they engage with AI shifts dramatically. They begin to set precise objectives and ensure the context is well defined. Many complaints regarding the inadequacies of AI often stem from users feeding in generic queries and receiving equally unremarkable responses.

The individuals who achieve optimal results view AI as a skill to master rather than a mere shortcut

Surprisingly, the strongest predictor of success with AI lies not in technical prowess but in a person’s mindset toward it. Those who approach AI as a skill worthy of development, rather than considering it a black box that either works flawlessly or fails miserably, tend to excel. The most adept users are those who actively engage in daily experimentation, constantly reflecting on how to secure better outcomes with each interaction. The goal should be to empower machines to serve our agendas rather than allowing them to think on our behalf; hence the need for a proactive, critical, and engaged usage of AI.

AI, like people, thrives on direction, feedback, and corrective guidance

Harnessing AI effectively taps into skills many individuals already possess: clear communication and thoughtful delegation. Just as you wouldn’t assign an intern a task and vanish, you need to engage with AI continuously. This includes breaking down complex projects, checking in frequently, and making necessary adjustments. The same principle applies; as the leader, you retain responsibility for the AI’s output, epitomizing the essence of ‘human in the loop’. It’s imperative to guide the AI constructively, ensuring the results meet appropriate standards.

Outsourcing your judgment to AI is a dangerous gamble, particularly when handling sensitive information

Not long ago, a manager from a small retail chain showcased to me an HR dashboard he’d created using AI. While impressive, he had unknowingly imported sensitive data, overlooking the ramifications of a potential data breach and the necessary compliance policies. I promptly directed him to consult with IT regarding the situation.

The concerns surrounding AI don’t stop at security risks. The models are shaped by human-born data that carries our intrinsic biases. When it comes to subjective assessments, such as deciding on a candidate’s suitability for an interview, the potential for bias escalates. Instead, concentrate on objective evaluations, such as verifying if candidates meet essential criteria like years of experience.

Ignoring the existence of AI will not mitigate its far-reaching implications

The growing environmental, ethical, and societal implications of AI cannot be ignored. I recently engaged with a charity focused on environmental preservation, where a director was caught between leveraging AI for advancement and grappling with the ethical concerns linked to its operational carbon footprint. Ignoring AI is not an option; it’s an established force. Cultivating a society that is well-versed and competent with AI will empower citizens to demand responsible and democratic practices surrounding its development. AI doesn’t merely represent a future opportunity; it’s currently in motion, and the pressing concern is who ultimately controls its trajectory.

The rapid advancement of AI demands swift and decisive action

What we’re currently witnessing in AI is merely the beginning; its capabilities are evolving at a pace most people are unaware of. An array of tasks that were confounding last year have transformed into routine functions. Where I once labored late into the night, troubleshooting code, I can now create entire applications in just hours with the simplest of prompts. Skepticism prevailed among developers when Anthropic’s CEO claimed that AI would soon write 90% of code. Today, many developers recognize the truth in that statement.

Unlike previous technological revolutions, the speed of AI advancement is eclipsing our ability to adapt. Historical transitions, like the movement from steam engines to locomotives, unfolded over decades. By contrast, today we witness a mere few months between breakthrough innovations and widespread adoption. We no longer have the luxury for protracted discussions; it’s vital we establish a societal and democratic framework that adapts as swiftly as technology, lest we find ourselves governed by tools we do not yet fully comprehend.

The individuals who will sculpt the manner in which AI molds our world need not possess technical backgrounds. It’s the explorers, those willing to acknowledge both its capabilities and its risks, who will take the lead. We all share the responsibility not only to grasp AI ourselves but also to advocate for its ethical use within our organizations, communities, and governments, ensuring we do not leave anyone behind in this transformational era.

Tom Hewitson is the founder and chief AI officer of General Purpose, an AI training company based in London.