Artificial intelligence promises extraordinary benefits. But in just a few years, how much of our world might we be willing to hand over to it?

I was born in 1955.

The world I entered would be almost unrecognisable to a young person today. There were no personal computers, no mobile phones, no internet, no email, no Google, and certainly no social media. Television itself was still something of a novelty in many parts of the world.

Communication was by letter or landline telephone. Photographs were taken on film and developed days or weeks later. If you wanted to know something, you consulted a book or encyclopaedia, visited a library, asked someone who might know, or simply accepted that the answer was not immediately available.

In the seventy-one years since then, I have watched technology transform almost every aspect of our lives.

And yet I now wonder whether the greatest transformation has only just begun.

Artificial intelligence has arrived with remarkable speed. Only a few years ago, the idea of holding a detailed conversation with a computer — asking it to analyse information, improve a piece of writing, explain a complicated subject, draft a complex letter, generate an image, or help solve a problem — would have seemed extraordinary.

Today, millions of people do precisely that every day.

So let us move forward a little.

Not fifty years. Not even twenty.

Imagine waking up on an ordinary morning in October 2030.

An ordinary day in 2030

Perhaps your personal AI assistant has already been working while you slept.

It has reviewed your emails and identified the three that actually require your attention. It has noticed that your electricity bill is unusually high and has compared alternative suppliers. It has checked your calendar, rearranged an appointment because of a scheduling conflict, and reminded you that your car is due for servicing.

Being the start of the week, it has used your personal preferences to produce your meal plan, compiled your grocery shopping list, checked current prices with your three preferred supermarkets, and placed the order with the supermarket offering the best value for money. Your order will be delivered by an autonomous vehicle or drone at 10:00.

It has also analysed information from your smartwatch and noticed a small but persistent change in your resting heart rate. It regards this as a routine matter and has scheduled a virtual consultation with an AI-approved virtual doctor at 10:30. The consultation will be fully funded by Medicare.

Your virtual consult identifies that you need to see your GP and your AI assistant schedules the appointment for the following day. Your doctor, when you see him or her, is also using AI. Before you enter the consulting room, a system has reviewed your medical history, medications, recent blood tests, and perhaps thousands of relevant medical studies. It has identified several possibilities for your doctor to consider.

None of this seems particularly far-fetched.

Indeed, much of the technology required to do it already exists in some form.

By 2030, the extraordinary thing may be that it will no longer seem extraordinary at all.

The enormous promise

It is easy to focus on the dangers of artificial intelligence, but doing so can obscure its enormous potential for good.

Medicine may be one of its greatest beneficiaries.

AI systems can already identify patterns in enormous quantities of information far more quickly than any individual human being. Applied responsibly, that capability could help doctors detect cancers earlier, identify rare diseases, design new drugs, and tailor treatments to individual patients.

Scientific research could accelerate dramatically. Problems that once required years of analysis might be tackled in months, weeks, or even days.

Education could also change profoundly.

Imagine every child at school having access to his or her own knowledgeable and understanding tutor, capable of adapting a lesson precisely to that child's abilities. A student struggling with mathematics might receive a completely different explanation from one who has already mastered the basics. Language barriers could largely disappear.

For older people, AI might provide another form of independence. Systems could help manage appointments and medications, identify potential scams, simplify increasingly complicated bureaucratic processes, provide local autonomous transport, and allow people to remain safely in their own homes for longer.

And in the workplace, countless hours of repetitive administration could disappear.

That could free human beings to spend more time doing the things machines do less well — creating, caring, persuading, leading, imagining, and simply interacting with other people.

That is the optimistic future.

But there is another possibility.

What happens to work?

Every major technological revolution has changed employment.

Machines replaced much agricultural labour. Factories transformed manufacturing. Motor vehicles transformed transport. Computers eliminated some occupations while creating entirely new industries.

Artificial intelligence is already beginning to do the same.

The difference is likely to be the speed and breadth of the change.

For much of history, automation primarily threatened physical labour. AI increasingly reaches into areas we once assumed required uniquely human intelligence.

Accountants, lawyers, programmers, designers, translators, journalists, administrators, financial analysts, and many others may discover that significant parts of their work can be performed by machines.

My current passion — the world of books and publishing — will not be immune.

That does not necessarily mean that entire professions will disappear. Some jobs will undoubtedly disappear, while others we cannot yet imagine will emerge.

The economic benefits could be enormous. So could the disruption.

If AI allows a business to accomplish with ten employees what previously required fifty, productivity rises dramatically.

But what happens to the other forty?

That question is not ultimately technological.

It is social, economic, and political.

Can we still believe what we see?

There is another problem that may be even more difficult.

For most of my life, a photograph provided reasonably persuasive evidence that something had happened.

A recording of someone's voice meant that the person had probably spoken those words. Video was more convincing still.

That assumption is disappearing.

AI can already create realistic photographs of events that never occurred, reproduce people's voices, and generate increasingly convincing video.

By 2030, we may routinely encounter material that is virtually indistinguishable from reality.

Imagine an election campaign in which a convincing video appears to show a candidate saying something outrageous.

Imagine a telephone call from someone who sounds exactly like your son or daughter telling you that they are in trouble and urgently need money.

Imagine fabricated evidence appearing during an international crisis.

The problem may no longer simply be that people believe things that are false.

It may also be that they stop believing things that are true.

A society in which nobody can readily distinguish reality from fabrication faces a problem far deeper than misinformation.

It faces an erosion of trust itself.

Who is watching whom?

AI also has the potential to know an extraordinary amount about us.

Our phones already know where we travel. Search engines know what interests us. Retailers know what we buy. Social-media platforms know what attracts our attention.

Now imagine artificial intelligence combining and interpreting that information.

Used benignly, this could be extremely useful. A system that understands our preferences could make our lives easier.

But the same capability could be used for surveillance, manipulation, or control.

Governments might employ increasingly sophisticated AI systems to monitor people. Companies could understand our behaviour — perhaps even our vulnerabilities — better than we understand them ourselves.

Privacy has been gradually eroded throughout the digital age.

AI could accelerate that process enormously.

What about war?

There is another area in which the consequences of artificial intelligence could be profound — warfare.

Wars have always been shaped by technology. Gunpowder, the machine gun, aircraft, radar, missiles, and nuclear weapons all changed not only how wars were fought, but also the balance of power between nations.

AI may represent another such change.

The current wars involving Russia and Ukraine, and the United States, Israel, and Iran, have already given us glimpses of how warfare is changing.

By 2030, increasingly autonomous drones might identify and track targets with limited human involvement. Swarms of relatively inexpensive unmanned aircraft could overwhelm weapons systems costing many times more. AI might analyse satellite images, intercept communications, identify patterns in enemy movements, defend computer networks, or launch cyberattacks at speeds no human command structure could match.

It might also save lives. Robots and autonomous vehicles could undertake dangerous reconnaissance, clear mines, deliver supplies under fire, or perform tasks that would otherwise put soldiers at risk. More accurate intelligence and targeting might, at least in theory, reduce civilian casualties.

But there is an uncomfortable question at the heart of all this.

If autonomous weapons reduce the immediate human cost of going to war — particularly for the country deploying them — might political leaders find military action easier to contemplate?

The arrival of nuclear weapons forced humanity to confront the consequences of possessing destructive power on an unprecedented scale.

Artificial intelligence may present a different challenge.

This time, the danger may lie not only in the power of the weapons we create, but in how much control over those weapons we are prepared to surrender.

From assistant to decision-maker

There may, however, be an even more important development.

At present, most of us think of AI as a tool.

We ask it a question. It responds.

We ask it to perform a task. It assists us.

But what happens when we begin allowing AI systems to act independently on our behalf?

Instead of merely suggesting a holiday, your AI could plan it and book it.

Instead of explaining different insurance policies, it could choose one and purchase it.

Instead of helping a company analyse its business, it could negotiate with suppliers, recruit employees, move money, commission advertising, and make operational decisions.

In medicine, an AI might move from assisting a doctor with diagnosis to recommending — or perhaps eventually initiating — treatment.

That transition may prove more significant than simply making AI more intelligent.

The crucial question may not be:

How clever can artificial intelligence become?

It may be:

How much authority are we prepared to give it?

And what if it becomes more intelligent than us?

Beyond today's AI lies a much more uncertain possibility.

Researchers sometimes speak of artificial general intelligence — AGI — meaning, broadly, a system capable of performing intellectual tasks across many areas at or beyond human level.

Beyond that lies the still more speculative idea of artificial superintelligence — ASI — intelligence substantially greater than our own.

Nobody knows whether such systems will be developed, or when.

Perhaps 2030 will arrive and today's AI will simply have become a much more capable tool.

Perhaps progress will slow.

Or perhaps something genuinely transformative will occur.

The uncertainty itself is important.

For the first time in human history, we are seriously contemplating the possibility of creating something intellectually more capable than its creators.

That should inspire both excitement and humility.

The question may ultimately be about us

Technology itself does not determine what kind of society we create.

Human beings do.

Nuclear physics gave us both nuclear medicine and nuclear weapons.

The internet gave us instant access to much of humanity's accumulated knowledge — and also scams, cybercrime, misinformation, and new forms of exploitation.

Artificial intelligence will almost certainly be similar.

It may help cure diseases while contributing to entirely new forms of illness or dysfunction — perhaps even changing the way our brains develop and operate.

It may liberate people from tedious work while depriving others of meaningful employment.

It may democratise knowledge while concentrating enormous power in the hands of governments or a small number of corporations.

It may make us more productive while making us increasingly dependent upon machines.

The technology will matter enormously.

But so will the choices we make about how it is used.

October 2030

October 2030 is not far away.

God willing, I will be seventy-five.

Perhaps I will look back at this article and smile at how much I got wrong.

Perhaps the changes will have been slower than expected.

Or perhaps I will be astonished at how conservative these predictions were.

What seems much less doubtful is that artificial intelligence will play an increasingly important role in our lives.

The challenge will be to embrace what it can offer without surrendering things we discover, too late, were worth preserving — privacy, trust, human judgement, meaningful work, personal responsibility, and genuine relationships with other human beings.

We have spent centuries inventing tools to extend human physical ability.

We are now creating tools capable of extending — and perhaps eventually surpassing — aspects of human intellectual ability.

That is an extraordinary achievement.

It is also an extraordinary responsibility.

Perhaps the most important question we will face between now and 2030 is not whether artificial intelligence can become more like us.

It is whether, in embracing it, we can preserve the things that make us human.