It’s interesting what people imagine when we think about “artificial intelligence”. I have personally seen everything from robots and android automatons to computer-generated imagery of circuits flowing through conduits, almost like something out of Tron, the 1982 sci-fi movie. For a lot of people, that is what AI looks like because that is what decades of movies, television and fiction have told us it looks like.

Where we are right now is considerably less dramatic. The systems we use today, such as ChatGPT and other large language models, are extremely sophisticated, but they are still largely driven by an input and a user. You ask something, the system analyses what you are asking, interprets the context and generates an output, and you continue refining that output until you are satisfied with the result.

A lot of this happens so quickly and so efficiently that it looks like magic, which I think is part of what creates the confusion, but in most normal interactions when the operator steps away from the device and stops prompting it, nothing further happens and the conversation effectively ends.

A lot of the fear taking up valuable thought space now is around what could happen with this technology. Sometimes those concerns are legitimate, such as criminals or foreign states using these tools, misinformation, security issues or the technology simply getting things wrong, but often the fear is also brought on by our experiences with fiction.

Think about WarGames, The Terminator or The Matrix. These movies, through their artistic portrayal and enormous production value, created very memorable scenarios where technology becomes autonomous, humanity loses control and everything goes horribly wrong.

Those stories are designed to be memorable, but a utopian society where nothing particularly bad happens and everybody becomes 20 per cent more productive does not make an interesting movie.

That does not mean there are no risks associated with AI. It simply means that some of our perception of those risks has been coloured by decades of entertainment that was specifically designed to create fear, tension and conflict.

What we have today is, in many ways, much more useful and much less cinematic. Large-language models, agentic LLMs and the various plug-ins, integrations and tools being built around them are nothing short of a modern-day miracle in terms of productivity. They can greatly increase work output and act as a force multiplier in a number of different ways.

Our company actively uses these tools to get more things done and to maximise efficiency across departments, but we try hard not to create artefacts, reports or documents simply for the sake of creating them. There is a good joke about LLMs that captures the strange phase we are in: one person feels obligated to make a report and says, “ChatGPT, make this report ten pages”, and the person receiving the report then says, “ChatGPT, review this ten-page report and summarise it for me.”

We have managed to take an incredibly powerful productivity tool and, in some cases, use it to create more work on both sides of the conversation.

This type of experience is unfortunately quite common right now because people are still learning how to use these systems properly, and there is an enormous temptation to generate more, because generating more is now easy. We are also in a phase where there is an immense amount of what people are calling “AI slop”, which is essentially low-quality and often purposeless content that can be generated very quickly and serves a fleeting purpose.

Think of one of those videos where a cat is driving a shopping cart down a highway for some reason. The ability to generate something instantly does not necessarily mean it is worth generating. That distinction is going to become increasingly important as these tools become more widespread.

A more serious component to the fear surrounding AI is the concern over what it could do to jobs and entire industries. What I can say with a reasonable degree of confidence is that there are going to be jobs created from this technology that we cannot properly fathom yet, while at the same time there will be jobs and industries affected by it.

We have gone through this before.

In the early 1900s, when automobiles started appearing in major European and American cities, there was enormous resistance from people whose livelihoods depended on horses. As uptake of automobiles increased, the use of horses for transportation began a long decline. For somebody living through that change, particularly somebody who bred horses, maintained stables or worked in one of the many professions supporting horse-based transportation, this probably looked close to an apocalypse, because their entire economic model was being challenged by a machine.

The difficult part is that they could very clearly see what was being lost, but they could not possibly see everything that was about to be created.

In 1910, you could get on a horse and ride somewhere with comparatively few supporting industries around the act of travelling. Now think about what it takes to support the modern automobile: raw materials, material sciences, engineering, design, manufacturing, metallurgy, vehicle inspections, insurance, gas stations, tyre shops, mechanics, driving schools, road construction, traffic management, dealerships, financing, rental companies, mobile windshield repair companies and probably another dozen industries that most people never even think about.

The automobile did not simply remove the horse. It created an entire economic ecosystem around itself. The people on the receiving end of that disruption could not reasonably have been expected to predict all of it. If we could travel back in time and ask somebody working in the horse economy what the future looked like, they would naturally have focused on the thing immediately threatening their livelihood rather than industries and professions that had not yet been invented.

This has happened many times, including within living memory. There was once the concept of the elevator operator, a person whose primary job was operating an elevator because elevators themselves were sufficiently complicated that people needed somebody there to operate them. As the technology improved, that vocation was relegated to history.

Travel agencies, although still around, are dramatically less prominent than they once were. They used to exist in specialty stores where people would physically sit down with somebody to book a flight or organise a vacation, while most of us now do a significant portion of that ourselves online.

Calculators were developed to help people solve maths problems without doing everything on paper. Calculators then moved into computers and smartphones, and now you can ask Google's Gemini, OpenAI's ChatGPT, or another LLM, to solve the problem, show you how it solved the problem and then explain it again in a different way if you still do not understand it.

In 1986, teachers protested the use of calculators in elementary schools over concerns that students would become dependent on the technology instead of learning basic mathematics.

Mathematics did not become useless because the calculator was invented, and people did not stop travelling because travel agencies declined. The tools and the economic structures around those activities simply changed, which is much closer to what I expect we are going to see with AI.

There will be efficiencies that certain professions go through, there will likely be some industries that contract in particular areas, and there will almost certainly be pressure on some junior roles where a meaningful part of the work is repetitive.

There is no point pretending otherwise, but there is also an enormous amount of upside that is much harder to see because a great deal of it has not been invented yet.

We are currently living through what I think is one of the most interesting periods in modern history because anybody with access to the internet can now access capabilities that were previously expensive, difficult to obtain or simply unavailable to them.

If you are looking for help with homework, you can ask an LLM to help you, and more importantly, if you word the request correctly you can ask it not to do the work for you but instead show you how to do the work yourself.

If you want to create a video game, you can ask where to start, what tools you need, how a game engine works and how to write the code. These tools can help you move from something that exists only in your head to a working prototype in hours or days instead of months or years, potentially with very little initial investment.

You can generate content, write a book, create animations, learn a language, analyse documents, solve engineering problems or simply use the system to help you organise your own thought process. The number of applications is almost impossible to list because the technology is general enough that people continue finding new ways to use it.

That is where I think the conversation becomes particularly important for Bermuda.

The biggest risk to Bermuda today is probably not AI itself, but lacklustre adoption of AI and LLM tools by the population and by businesses here. We are a small jurisdiction with a limited labour pool, and there are many specialist skills which are expensive or difficult to obtain locally.

Small businesses in particular cannot simply hire another five or ten people every time they want to expand into another capability, which means tools that allow one person to become significantly more productive should be extremely interesting to us.

In our own company, we have helped dozens of people install ChatGPT on their phones and shown them different ways of using it usefully. Sometimes that is the hardest part because people often barely scratch the surface of what it can do, a little like buying a computer and using only the calculator.

The real value starts to appear when you give the system context, ask follow-up questions, challenge its answers, ask it to explain something differently, or tell it exactly what you are trying to accomplish. These systems are not perfect, and they make mistakes, hallucinate and can produce nonsense with confidence, so proper use still requires human judgment and some understanding of the subject matter.

There will continue to be a lot of discussion around what AI may become in ten, 20 or 50 years — and there probably should be. But right now, the reality is that millions of people suddenly have access to tools that can help them learn faster, work faster, build things faster and gain access to capabilities that were previously beyond their reach.

For a place like Bermuda, I think the much bigger mistake would be watching everybody else learn how to use these tools while we sit on the sidelines worrying about The Terminator.

• Zlatko Zahirovic is the cofounder and chief technology officer of Paradise Mobile and deputy chairman of the GSMA North America Terminals Working Group