It’s the Year 2030. What Will Artificial Intelligence Look Like?

The Wall Street Journal asked a selection of experts from academia, business, consulting firms and think tanks to weigh in on what AI will be doing in 2030. Below are some predictions.

A gradual, but profound change

Artificial intelligence is advancing rapidly. Some experts suggest we may achieve artificial general intelligence—machines that can outperform humans in virtually every task—as soon as 2033. The implications are profound, potentially reshaping industries, economies and the very nature of work.

But there’s a crucial disconnect between technological advancement and mass adaptation by organizations and society. As scientist and futurist Roy Amara famously observed, “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.”

The year 2030 will likely put us right between the short- and long-run implications of AI.

So despite AI’s lightning-fast technological progress, we shouldn’t expect to see immediate, sweeping global effects in the next few years. The integration of AI into our daily lives, workplaces and institutions will be gradual, as these things change much more slowly than technology does.

As we navigate this dual-speed reality, we must prepare for a future where AI’s long-term effects surpass our current imaginations of what it can do—even as its short-term influence may fall short of the most ambitious predictions.

—Ethan Mollick, professor of management at the University of Pennsylvania’s Wharton School 

AI as smart as humans? Not likely. 

Researchers and entrepreneurs often say that we will reach artificial general intelligence—machines that can do essentially any of the cognitive work that humans can—before the end of the decade, or even within a couple of years.

I seriously doubt it.

To begin with, there is the rampant tendency of the current systems called large language models to hallucinate and make stupid errors; worse, there is no principled solution in sight to solve these problems. Only hope. Hope that more data and more powerful processing chips will magically fix all the issues.

But we are running out of the fresh, valuable data that are needed for LLMs to improve. They already have swallowed up nearly the entire internet. Now they are starting to choke on their own hallucinatory fumes. From 2020 to August 2022 there was rapid progress in the power of LLMs, but already that has slowed. We need genuine innovation, and that takes time.

From a technical standpoint, 2030 may be just like 2023, with better graphics. Behind all that, there is a risk that the AI investment bubble may soon burst, which will leave many people rushing for the exits; it may take years to regroup.

Anticipating artificial general intelligence by 2030 is just not realistic.

—Gary Marcus was a founder and chief executive of a machine-learning company that was sold to Uber and is the author of “Taming Silicon Valley”

Transformation, not job loss  

Here’s what happened in the second half of the 2020s: AI advances fueled the most profound business transformation in history. By 2030, AI systems reached unprecedented levels of capability, reshaping industries and jobs alike. While fears of widespread joblessness persisted, reality unfolded differently—transformation, not job loss, defined the era.

The reason lies in the nature of business evolution. Even groundbreaking technologies—in fact, especially groundbreaking technologies—require time and effort to translate into productivity gains. New skills, processes, and business models must be invented. Fortunately, enough businesses were up to the challenge and as a result, the late 2020s witnessed a surge in productivity.

Yet the transformation was uneven. Over half the Fortune 500 vanished, replaced by a wave of new titans, including an unprecedented number of trillion-dollar enterprises. Occupations evolved similarly—some went the way of the elevator operator, while new ones emerged. The set of occupations and tasks affected were quite different from those of earlier transformations. This time, creative workers, professionals, writers, managers and programmers were among the most affected.

The successful companies all had one thing in common. They didn’t simply buy into AI hype. Instead, they methodically applied a task-based approach. This recognizes that tasks—not jobs, products, or skills—are the fundamental units of organizations. AI revolutionized certain tasks like coding and customer service, while others remained relatively unaffected. 

As of 2030, humans still retained an edge in handling situations where there was a lack of historical data or structured rules available. Our ability to improvise outperformed machines.

—Erik Brynjolfsson, director of the Stanford Digital Economy Lab and co-founder of Workhelix

AI everywhere, like the internet

The biggest difference between AI in 2030 and AI now will be its degree of integration into companies. In the early 2000s, another rapidly evolving general-purpose technology—the internet—was just being broadly adopted. Organizations that bolted it onto existing structures saw some benefit, but those that built from the ground up around internet access soon became the world’s most valuable companies.

 Something similar will likely happen with AI. By 2030, we’ll see many organizations—some new, some radically transformed—with AI embedded in their structure. Every employee will access it regularly and seamlessly: to bounce ideas off it, to manage and automate tasks and to get feedback about a company’s services or products. Even by 2030, such organizations will be in the minority, but their productivity gains will highlight the value of this approach.

Only when AI stops being seen as a new technology—and becomes just an assumed part of everyday work and life like the internet—will its true potential be realized. The year 2030 will mark the midpoint of that shift.

—Alex Singla, global leader of QuantumBlack, AI by McKinsey

AI personal assistants to handle our lives

By the year 2030, we will each have a Personal Large Action Model, or PLAM. These are advanced AI agents designed to replicate and emulate our unique decision-making processes.

Today, a system like ChatGPT does its best to approximate how individual people speak and write using limited data; in the future, a PLAM will use data collected from the devices we wear (earbuds, continuous glucose monitors) and the devices we use (smart toilets embedded with sensors, digital wallets) to understand our likely behaviors and act on behalf of us.

 Our PLAMs, which we will train through repeated use, will learn and adapt to our individual unique behaviors, actions, mood expressions and preferences, and they will invisibly and autonomously handle complex tasks, such as negotiating rates with CLAMs, or Corporate Large Action Models. Here’s how that could look: As a business traveler, you have a coming trip from New York City to Munich, and there are no direct flights. Your PLAM would let airline carriers know it wants a deal and allow them to compete for your business.

After brokering an itinerary optimized to what you specifically like (preferred seating configuration, just the right amount of time to make your connection, ability to accumulate points), your PLAM completes all of the tedious steps involved in actually purchasing the ticket: signing in to the airline and authenticating, entering your information, selecting which credit card to use, authenticating for payment, submitting payment details, entering the receipt details in your company’s accounting system and posting travel details to your calendar. With enough PLAMs deployed, yours could work to find you the perfect seatmate, too.

—Amy Webb, chief executive, Future Today Institute

AI agents as collaborators at work, too

The advent of “AI agents” will be the turning point for artificial intelligence in the next five years. AI agents trained on what is highly relevant to the user, both professionally and personally, will protect us from receiving email, phone calls, texts and instant messages that aren’t of much use to us. These agents will handle this automatically—possibly through negotiations with other people’s agents—to keep most of this communication at our periphery. They will respond to these messages, organize them and push away those that aren’t germane.

Today, after a meeting, Microsoft’s AI tool called Copilot will give me a summary based on what was discussed—but future AI agents will know what is really important to me.

Rather than simply automating tasks, AI agents will become true collaborators. It will be the difference between yet another tool to enhance our everyday tasks, and the advent of true technology companions that will transform the way we work, buy, learn and interact.

—Erick Brethenoux, artificial intelligence chief of research, Gartner 

‘Empathy’ bots for sale to kids

By 2030, it will be commonplace to use AI emotional companions, not just for romance, therapy and eldercare, but also to provide love and “empathy” for children and teens. 

Currently there are bots that educate and entertain children, but nothing like the huge upsurge of intimate-relationship bots we see in adults. Yet by 2030, the market opportunity businesses see in loneliness will accelerate the development and dissemination of the child version of love or “empathy” bots. For younger kids, there will be physical “buddies,” perhaps taking the form of cute Teletubby-like robots with expressive faces. Older kids and teens may wear a smart device that is “always listening” and providing frequent validation.

But there are serious risks to replacing early experiences of mutual human empathy with fake bot” “empathy.” In real-life relationships, children encounter disagreements and challenges through which they recognize that other people have their own actual feelings and perspectives. This is the engine for developing mutual empathic curiosity, which is crucial for successful adult relationships. In contrast, bots, even if programmed to provide some disagreement or questioning, can only provide one-way simulated “empathy,” passively received by the child. The child cannot learn about the authentic inner life of another human. 

It isn’t just that bot “empathy” is fake and unidirectional. Adults develop strong attachments to their relationship bots, making them vulnerable to dangerous manipulation. With profits depending on capturing the user’s attention, tech companies already use irregular rewards, such as social-media likes, to create addiction. Children are more susceptible. Given that childhood is the critical time to develop one’s capacities for relationships, stealing attention away from real life is likely to have much more serious consequences.   

—Jodi Halpern, Chancellor’s Chair and professor of bioethics, University of California, Berkeley

Autonomous robots gain independence

By 2030, AI will greatly enhance the capability of robots to function independently in complex environments.

Robots will reach advanced levels of autonomy, making high-level decisions with minimal human input while collaborating and learning from each other. Through cloud communication, new capabilities and data will be easily shared among multiple robots, enabling them to work together on a task.

With enhanced agility, collaboration, adaptability and dexterity, autonomous robots will better navigate difficult environments such as off-road terrain, forests or cities. Robots will also perform tasks in human environments like homes and hospitals, such as cooking, helping the elderly, serving food or cleaning in buildings.

Human-robot interaction will improve, with robots communicating more naturally and understanding human emotions and intentions.

Robots’ new abilities will boost productivity, streamline operations and enhance safety, transforming how work is done. For instance, ground and aerial robots that collaborate with each other and with humans will provide aerial delivery and search and rescue after natural disasters. Automation will have a particularly big impact on sectors like manufacturing, logistics, construction, security, healthcare, transportation and space exploration. 

The shift will create new human roles in robot oversight, AI system design, and maintenance, but it may also require significant workforce reskilling. Ensuring the safety and reliability of AI-driven robots will be crucial; when an AI system is embodied in a robot, faulty actions could lead to damage in the real world. This highlights the importance of robust safety mechanisms and thorough testing to ensure AI-driven robots operate securely. 

—Prof. Giuseppe Loianno, director of the Agile Robotics and Perception Lab at New York University

The AI-powered doctor’s office 

By 2030, AI will be deeply integrated into healthcare, transforming how patients and providers interact. 

AI tools that quickly analyze lab results and scans will help speed detection and diagnosis of conditions like cancer or heart disease. Systems that combine different types of data, like images, genetic information and medical records, will give doctors a more complete picture of a patient’s health, leading to better diagnosis and treatment. Critical conversations about diagnoses and treatment will remain the responsibility of healthcare professionals, but AI will offer support. For example, AI will ensure patient information is customized to their age, education level and health condition. 

While human empathy and judgment will still be critical in delivering the best care, AI will help nurses by monitoring patients’ vital signs. It will suggest treatment plans based on a patient’s medical history and warn doctors about possible problems like worsening symptoms or harmful drug interactions. AI will also handle time-consuming paperwork by automating tasks like coding, billing, and managing electronic health records. All this will reduce stress for doctors and nurses, enable them to spend more time with patients, and lead to quicker, more personalized care.

Even when patients aren’t at the doctor’s office, wearable devices will be able to monitor more disease indicators and send real-time information to AI systems that alert patients and their doctors about potential health risks. This will shift healthcare toward preventing illnesses, rather than just treating them after they occur.

With AI’s use, ethical and legal concerns will become more important, and concerns about safety and fairness will be addressed by clearer rules, monitoring and education.

—Metin N. Gurcan, director of the Center for Artificial Intelligence Research, Wake Forest University School of Medicine

Societal trust is at risk

AI-generated content in 2030 will be far more realistic than it already is, further blurring the boundaries between reality and fiction. The mere existence of such high-quality, AI-generated content will give people, including politicians, cover to question the truth—even if AI wasn’t used. 

In August, Donald Trump claimed that a photo showing the size of the crowd at a rally for Vice President Kamala Harris was AI-generated, an assertion media outlets said was unfounded based on photos and video of the event. 

Without decisive action, we may be operating in a bifurcated reality, where anything can be true or false depending on whether an individual is already primed to believe it. In this dystopian future, truth is subjective, and reality is whatever reinforces one’s prior beliefs.

What is certain is that over the next five years, AI-generated media is likely to become only more realistic and more pervasive. Without widespread education about the danger, the shared reality and an informed public on which democracies so depend may be at existential risk. 

—Valerie Wirtschafter, a fellow in the Foreign Policy, Artificial Intelligence and Emerging Technology Initiative at the Brookings Institution

Elizabeth Seay, a features editor for The Wall Street Journal in New York, contributed to this article. Bart Ziegler is a former Wall Street Journal editor. They can be reached at reports@wsj.com.

Artificial Intelligence

Read the full report

It’s the Year 2030. What Will Artificial Intelligence Look Like?

The Joys of Talking to ChatGPT

Will AI Make Job Recruiting More Efficient—but Less Fair?

So, You Think You Know ChatGPT? Take Our Quiz

Can AI Companions Help Cure the Loneliness Epidemic?

7 Everyday Work Problems AI Helps Me Solve

The Secret to Talking to an AI Chatbot

How AI Has Already Begun to Change These Workers’ Jobs