After ChatGPT and GPT-4, the disruptive potential of artificial intelligence is impossible to ignore. Here’s how to navigate the changes now under way.

The future is here and about to get much more evenly distributed. After years in the laboratory and on the movie lot, artificial intelligence is real, and ready for its close-up.

Unless you’ve been living under a rock, you have heard of ChatGPT and GPT-4. You have probably played around with it. Your kids certainly have. So, what now? Anybody can see that AI has become incredibly powerful. To imagine that it does not change everything is simply a failure of imagination. But how, precisely, is it going to impact our work, our lives, our world? And how can it be leveraged for positive outcomes?

We are on the cusp of a moment of momentous change. A period which will fill the history books – or perhaps the history downloads or chip implants – of the future. A period in which your future is unwritten but depends on what you, with your corporate and personal hats on, do next.

In 2017, I co-authored a book with Malcolm Frank and Paul Roehrig, called What to Do When Machines Do Everything, which set out a corporate playbook for executives looking to leverage the incredible power of the nascent machine-learning techniques that were then just emerging. Now, six years later, with technology that has developed and is maturing fast, I offer this AI survival guide – a framework made up of six steps to help you and your organization (and your loved ones) think about how to navigate terra digitalis incognita.

1. Mapping

The territory is new and unknown and there are no published maps available, so you must create your own. Take a tip from the French in the Seven Years’ War and hire or commission scouts to head out into the woods and find what is there – what natural resources, what predators, what potential points of fortification. In 2023, that means establishing a handpicked task force of some of your best in-house technologists to explore the raft of AI technologies and products that are emerging daily (hourly, in fact), speaking with analysts and academics who have had early access to said tools, and contracting with consultants who understand your business processes and can align them against new relevant tools. It means feverishly reading everything from the MIT Technology Review to the subreddit on machine learning ( Lastly, and most importantly, it means walking the via regia of AI, going to the kings of this new world. Leaders should be knocking on the door – metaphorically, or even better, literally – of 3180 18th Street, San Francisco, California, to understand what OpenAI can do for them.

It is incumbent upon you to understand what is going on with AI. Now. Not next quarter. Now. Without a map, you are – and have – lost.

2. Developing

“Whoever becomes the leader in this sphere will rule the world.” Vladimir Putin’s ominous 2017 words, spoken in the context of geopolitical turmoil, are – uncomfortable as this may be – entirely applicable as a description of the competitive AI battles that exist between commercial organizations.

Whichever company becomes the leader in this sphere will rule the world.

However, while a small number of companies will undoubtedly have a leadership position in AI as it continues to develop (Google over recent years, Microsoft potentially in the next few), anti-monopoly competition laws will likely prevent any one company from completely controlling AI. Rather, several companies will be dominant players. If it is at all possible, your company should be one of them. In every previous phase of technological development, the developers of the primo tool of the age have stood as Colossi of all they survey; Whitney and the cotton gin, Watt and the steam engine, Stephenson and the locomotive, Mercedes, Benz, Ford and the automobile, Gates and the graphical user interface Now.

The surest way to be a master of the coming universe is to be a developer of AI technology – be it directly yourselves, in partnership with others, in consortia, or through investment funding. The development of large language models (LLMs), the heart of AI, will continue to be the purview of a small number of very specialized software engineering firms for the foreseeable future, but the integration and customization of those LLMs with other component technologies and processes – ones your company may have expertise in – will allow your technology to be a component of the broader, more complex ‘engine’ that AI will become.

This possibility will be open to many companies in many different technology areas (including user interfaces, digital avatars, integration, hosting, managed services) and many different business domain areas (including AI for wealth management, for medical diagnostics, for education, and so on.) The closer your organization can get to the source of proprietary leading-edge AI creation, the better.

3. Mastering

To paraphrase Kevin Kelly, the founding editor of Wired, the future of your work is “X + AI,” where ‘X’ is whatever it is that you do, and ‘AI’ is, well, AI. That simple motto captures a profound insight – that adding AI to your job as a coder, a doctor, a teacher, an executive, a chief executive – and to your product or service, as a car manufacturer, a bank, an insurance company, an online gaming company – is the route to the next performance threshold available to you.

In a previous Dialogue article called ‘Leeches are for suckers’ (Dialogue, Q3 2018) I outlined how managing greater and greater density and complexity of data at faster and faster speeds was the key task ahead. Now, mastering that key task is mandatory, not elective. You personally, and your organization corporately, may not develop AI tools, technologies, or techniques yourselves, but deploying and optimizing them within the context of whatever it is you do is set to be crucial in the competitive game you are playing. The legal firm that can process huge volumes of information and provide legal insights faster than its competitors will become the go-to legal firm. The retailer that can deduce fickle fashion sentiment and adapt to market demand faster than its competitors will become the next sensation.

As has been frequently noted, AI itself will not destroy jobs and organizations, but people and organizations using AI will destroy the jobs of people and organizations not using AI.

4. Zagging

Having been an AI enthusiast and evangelist for many years (my undergraduate dissertation was on ‘The Philosophy of AI’ – in 1983!) it is exciting to see AI finally break through into mainstream consciousness. But having been a Marxist contrarian – of the Groucho variety, not wanting to belong to a club that would have me as a member – for even longer, the explosion of interest sets my non-conformist alarm bells ringing. I am reminded of the Oscar Wilde quote: “Whenever people agree with me, I always feel I must be wrong.”

If everyone now thinks that the future is X + AI, it is probably not. The career and/or corporate manifestation of this impulse is to hedge against AI: to develop a job, product or service that is explicitly designed around what AI cannot do, currently or in the forecastable future. In essence, to zag when everyone else is zigging.

For instance, a hotel (and its employees) could double down on offering non-technology mediated ‘experiences,’ perhaps including ‘digital detoxes’. A department store could double down on non-technology mediated human-delivered customer service. A financial advisor could double down on the quality of its non-technology mediated advice and assurance.

Though AI will undoubtedly diffuse into more and more aspects of our work and lives over the next few years, we should not overlook that the AI software market – estimated to be worth $135 billion by 2025 – is a relatively small subset of the $4.5 trillion global IT industry, which itself is a subset of the entire $100 trillion global economy. There is still a lot of opportunity outside of the tech sphere, no matter how fast tech generically, and AI specifically, develop in the next few years.

5. Stopping

There is something of a ‘manifest destiny’ quality to AI – that it is a natural, irreversible dynamic set to unfold no matter what any of us think or feel about it, individually or collectively. But the truth is that AI, as a human creation, is still subject to human control. And within this truth is the reality that though AI has its proponents, it also has its detractors – critics who, for a variety of reasons, believe that we should pause, stop or, at the very least, legislate how AI is developed and deployed.

Sometimes these reasons are philosophical; the fear that AI could potentially overwhelm human control and lead us towards the ending of 2001: A Space Odyssey. (“I’m sorry, Dave. I’m afraid I can’t do that.”) Sometimes they are geopolitical; see Step Two, and the thoughts of President Putin. And sometimes they are commercial; the concern that a dominant technology could create a dominant vendor with an unsurpassable monopoly position.

As I type, the campaign for technological disarmament – call it CTD – has broken into public view. The Italian government has banned ChatGPT and a raft of senior technologists, academics, politicians, and investors, including Elon Musk, Steve Wozniak and Andrew Yang, have released an open letter calling for a moratorium on AI development, stating that the challenges AI raises to humans “must not be delegated to unelected tech leaders”. Perhaps you or your organization has a reason to stop AI development. Aligning with this nascent campaign may strike you as a Luddite move, but in fact it will be entirely rational. Human agency is still our most profound quality. The power to shape the future is, and should be, still in human hands.

6. Hiding

Sun Tzu’s advice on dealing with a stronger opponent was typically pithy: “If he is in superior strength, evade him.” For many organizations and people, the only real option will be to keep as far away from AI as possible. Stay small, stay analogue, stay off the grid, stay local, stay home, stay human. Though this thought may seem heretical from someone who has made a living in the tech industry for 40 years, it is born from the knowledge that many organizations and people simply will not be able to compete against machines, and resistance will be futile. Increasingly in my job I have the sense that I am a bad robot, and GPT-5, 6 or 7 will be able to do what I do, better than I do. Many people and companies will be able to leverage AI to augment their work – see Steps Two and Three – and reach a new promised land of productivity, fulfillment and wealth. But many will not.

A clear head will be required as to understanding the possibilities to ‘head out west’, or to stay put and make the best of where you are now. That map is going to come in handy.

The value of planning

As President Eisenhower once said, “Plans are useless, but planning is indispensable.” This is always good advice, but particularly at moments of existential change. Such is the moment now.

To survive – let alone thrive – in the age of AI will require agility, courage and good old fashioned human intelligence. Much is at stake. Bon chance, and bon voyage – and take this AI survival guide with you.

The views expressed are those of the author and do not represent any official position held by Gartner Inc.