Past predictions for 2026 vs. Reality

This article looks at what people once thought 2026 would be like. It compares predictions made 5, 10, 20, 50, and 100 years ago with what has actually happened by 2026.

1. 5 Years Ago — 2021

In 2021, AI was advancing quickly, remote work was becoming part of everyday life, and people were making detailed forecasts about what the middle of the decade might bring.

Prediction: AI Would Spend More Time Thinking

In August 2021, AI researcher Daniel Kokotajlo published an influential forecasting essay titled What 2026 Looks Like on LessWrong.

He presented it as a possible future history, not a guaranteed prediction. One part of his scenario imagined AI moving beyond simply making larger models and instead using more computation at inference time. He described possible systems as “bureaucracies” that would spend more time working through a problem before producing an answer.

Reality in 2026

By 2026, test-time reasoning and more agentic AI systems had become important parts of frontier AI development.

METR reported in February 2026 that newer evidence suggested developers were getting more benefit from AI than they had in its earlier 2025 study. METR also warned that selection effects made the size of the improvement difficult to measure reliably.

By May, METR reported that some of the most capable coding agents it evaluated could complete software tasks that would take human experts days or weeks. The agents still made bugs and poor decisions on harder tasks.^{[1] [2]}

OpenAI and Anthropic also continued developing systems that spend additional computation on difficult problems. Anthropic's Claude Opus 4.6, released in February 2026, introduced adaptive thinking and longer-running agentic capabilities. OpenAI's o-series had already established extended reasoning and tool use.^{[3] [4]}

Prediction: AI Would Become More Powerful at Propaganda

Kokotajlo's 2021 scenario also imagined that AI-powered propaganda would become much more powerful as models improved.

He described increasingly sophisticated systems that could study audiences and optimize persuasive material, creating a more distorted and polarized information environment.

Reality in 2026

AI-generated images, video, audio, and political advertising are now established parts of the information environment.

Reporting during the 2026 U.S. election cycle documented AI-generated campaign material and deepfakes, including altered or fabricated images of political opponents.^[5]

However, the available evidence does not show that AI persuasion has taken control of voter behavior or replaced existing political identities.

Prediction: A U.S.–China Fight Over Computer Chips

Kokotajlo's 2021 scenario also imagined that competition over advanced computing would lead to a major U.S.–China conflict over chips, including export controls and tariffs.

Reality in 2026

The United States imposed major controls on advanced computing chips and semiconductor-manufacturing equipment destined for China beginning in 2022.

In January 2026, the Bureau of Industry and Security changed its licensing policy to allow case-by-case review of some advanced chips, including NVIDIA H200 and AMD MI325X systems, while keeping security conditions in place.^[6]

By 2026, advanced computing technology had become an important part of the U.S.–China technology conflict.

Prediction: Remote and Hybrid Work Would Last

In February 2021, the McKinsey Global Institute estimated that 20% to 25% of the workforce in advanced economies could potentially work from home three to five days per week after the pandemic.

McKinsey described this as a potential structural change rather than a precise prediction about what the workforce would look like in 2026.

Reality in 2026

Remote work remains a substantial part of U.S. working life, although it did not replace office work.

WFH Research estimated that about 27% of paid full workdays in the United States were worked from home in March 2026.^[7]

The result is therefore close to the scale McKinsey identified as potentially feasible, but the workplace model is mixed rather than fully remote.

2. 10 Years Ago — 2016

In 2016, autonomous vehicles, the sharing economy, subscriptions, and new forms of transportation were attracting major attention.

Prediction: Fully Autonomous Ride-Hailing by 2021

On August 16, 2016, Ford announced plans to have a high-volume, fully autonomous SAE Level 4-capable vehicle in commercial operation by 2021 through a ride-hailing or ride-sharing service.

Ford said the vehicle would be designed without a steering wheel, gas pedal, or brake pedal.^[8]

Reality in 2026

Ford did not deliver the promised 2021 high-volume Level 4 vehicle.

At the same time, fully autonomous ride-hailing did become a real commercial service.

Waymo reported in February 2026 that it was operating fully autonomous ride-hailing in several metropolitan areas and expanding into Dallas, Houston, San Antonio, and Orlando. By April, it had opened Miami and Orlando to the general public.^{[9] [10] [11]}

Ford's own autonomous-driving program also changed course after its investment in Argo AI was impaired.

Scenario: “You'll Own Nothing and Be Happy”

In November 2016, Danish politician Ida Auken published an essay for the World Economic Forum titled “Welcome to 2030: I own nothing, have no privacy, and life has never been better.”

The essay described a hypothetical 2030 city in which many products had become services and residents owned very little.

A later author's note clarified that the piece was a scenario intended to stimulate discussion rather than a personal prediction or political program.^[12]

Reality in 2026

Subscription and access models expanded, especially in digital services. Physical ownership, however, remains widespread.

The U.S. Census Bureau reported a national homeownership rate of 65.0% in the second quarter of 2026.^[13]

At the same time, the European Union's right-to-repair rules began applying on July 31, 2026. The rules strengthen consumers' ability to have qualifying goods repaired rather than replaced.^[14]

The result points in both directions: service models grew, but ownership did not disappear.

Prediction: Supersonic Passenger Flights Would Return

In November 2016, Boom unveiled its early supersonic-airliner concept in cooperation with Virgin Galactic's engineering organization.

The proposed aircraft was designed around Mach 2.2 cruise speeds. The company was targeting passenger service during the 2020s, with early reporting putting scheduled flights around 2023.^[15]

Reality in 2026

Commercial supersonic passenger service had still not returned by September 2026.

Boom's XB-1 demonstrator first flew in 2024 and later broke the sound barrier. Overture, however, remained a development program rather than an operating passenger aircraft.

In 2026, AFWERX reported that Boom expected Overture flight testing to begin in 2026, with passenger flights planned for 2029. The FAA was also developing a new regulatory framework for civil supersonic flight.^{[16] [17]}

3. 20 Years Ago — 2006

In 2006, the web was rapidly expanding, smartphones were emerging, and people were trying to imagine what the connected world would become.

Prediction: A Low-Cost Global Mobile Network

In September 2006, the Pew Research Center's Internet & American Life Project published The Future of the Internet II, based on an opt-in survey of 742 internet stakeholders.

One of its central scenarios predicted that by 2020, worldwide network interoperability would be highly developed and mobile wireless communications would be available to anyone anywhere at extremely low cost.^[18]

Reality in 2026

The broad direction of the prediction was close to reality.

The GSMA's Mobile Economy 2026 reported 5.8 billion unique mobile subscribers, equal to about 70% of the world's population. It also reported 8.8 billion wireless connections.^[19]

The International Telecommunication Union separately estimated that about 6 billion people were using the Internet in 2025.^[20]

Mobile connectivity is now widespread, although universal access has still not been achieved.

Prediction: Immersive Virtual Reality Workplaces

The 2005 BT Technology Timeline, edited by Ian Neild and Ian Pearson, included forecasts for increasingly immersive digital environments.

These included holographic displays, virtual environments, and other technologies that could reshape work and daily life during the following decades.

Reality in 2026

Mixed-reality hardware became much more capable, but immersive virtual reality did not replace conventional office computing.

Counterpoint Research reported that global VR headset shipments fell 18% year over year in the second quarter of 2026, even while enterprise demand remained comparatively resilient.^[21]

Apple continued to promote Vision Pro for business uses such as design, training, and productivity. These uses coexist with ordinary computers and displays rather than replacing them.^[22]

Prediction: $1,000 of Computing at Roughly 10¹⁶ Calculations per Second

In The Singularity Is Near, Ray Kurzweil argued that roughly 10¹⁶ calculations per second would be enough to model the functional computational activity of the human brain.

He wrote that by around 2020, this amount of computing power could be available for about $1,000.

The forecast concerned computational capacity. It was not a guarantee that a $1,000 computer would possess human-level intelligence.^[23]

Reality in 2026

The exact $1,000 single-machine target has not been reached in the straightforward sense Kurzweil described.

NVIDIA's 2026 specifications list the RTX 5080 at 171 TFLOPS on its ray-tracing throughput specification, while its listed price is around $999. Meanwhile, the June 2026 TOP500 list contained five exascale-class supercomputers. The leading system measured above 2 exaflops on the HPL benchmark.^{[24] [25]}

The comparison is imperfect. Different operations, architectures, and benchmarks are not interchangeable. Raw FLOPS also cannot be treated as equivalent to the computational requirements of a human brain.

The safer conclusion is that available computing grew rapidly, but Kurzweil's specific $1,000-at-10¹⁶ target should not be considered achieved.

4. 50 Years Ago — 1976

In 1976, scientists, journalists, and futurists were looking ahead to a distant technological future. Some of their forecasts were surprisingly close.

Prediction: Solar Power Would Grow While Fusion Stayed Difficult

On July 4, 1976, Thomas O'Toole, science editor of The Washington Post, published “Inventing the Future,” a wide-ranging forecast of American technological life.

Among the ideas discussed were expanded solar energy, medical advances, and the continued difficulty of making nuclear fusion a practical energy source.^[26]

Reality in 2026

Solar power became a major source of new U.S. electricity capacity.

The Energy Information Administration reported in February 2026 that developers planned 43.4 GW of new utility-scale solar capacity for 2026. That represented 51% of all planned utility-scale electric-generating additions.^[27]

EIA also reported that solar supplied about 6.7% of U.S. utility-scale electricity generation in 2025.^[28]

Fusion still had not become a normal source of grid electricity. The Department of Energy's June 2026 roadmap targeted a U.S. fusion pilot plant in the mid-2030s.^[29]

Prediction: Nuclear-Powered Artificial Hearts and Longer Lifespans

O'Toole's 1976 forecast also discussed medical predictions involving nuclear-powered artificial hearts and longer lifespans.

The nuclear-heart idea was not fictional. U.S. researchers were actively developing plutonium-powered artificial-heart systems and studying whether they could be used clinically.^[30]

Reality in 2026

Nuclear-powered artificial hearts did not become routine medical technology.

Historical National Library of Medicine records show that radiation risks led to opposition to plutonium-powered artificial hearts for human use. Modern mechanical circulatory support instead uses electrically powered systems.^[31]

Life expectancy, however, increased substantially. The CDC reported in January 2026 that U.S. life expectancy at birth reached 79.0 years in 2024, compared with 72.8 years in 1976.^{[32] [33]}

Prediction: Population Growth Would Slow

In The Next 200 Years, Herman Kahn, William Brown, and Leon Martel argued that the world was approaching a turning point in population growth.

Their “surprise-free” scenario expected population growth rates to decline as economic and technological development spread. They also expected food production and economic output to continue rising.^[34]

Reality in 2026

The broad demographic direction was correct. Fertility has fallen across much of the world, and global population growth is slowing.

UN publications in 2026 continue to document major differences between high-fertility and low-fertility societies as the demographic transition continues.^[35]

The economic and poverty comparison is more complicated.

The World Bank's March 2026 update estimated that 10.4% of the global population lived below its revised extreme-poverty line in 2024. It also added nowcasted estimates through 2026.^[36]

In June 2026, the World Bank projected global growth of 2.5% for 2026 while warning that developing economies were still struggling to close income gaps with advanced economies.^[37]

5. 100 Years Ago — 1926

In 1926, 2026 was a century away. Yet some observers were already imagining technologies that would eventually feel surprisingly familiar.

Prediction: The Vest-Pocket Wireless Videophone

On January 30, 1926, Collier's Weekly published an interview with Nikola Tesla conducted by John B. Kennedy.

Tesla imagined a world in which wireless communication would allow people to communicate instantly across great distances. He also imagined being able to see and hear someone as though they were face to face.

He predicted that the equipment would eventually be simple enough to carry in a vest pocket.^[38]

Reality in 2026

The functional core of Tesla's prediction is remarkably close to modern communication.

The GSMA's 2026 mobile-industry report counted 5.8 billion unique mobile subscribers worldwide. The ITU estimated that about 6 billion people were online in 2025.^{[39] [40]}

Billions of people now use mobile devices connected to global networks that support real-time voice and video communication.

Prediction: Wireless Power Across Great Distances

In the same 1926 interview, Tesla also predicted that wireless transmission of power would eventually become commercial and transform energy transmission.

The prediction was broader than a literal description of a modern aircraft being powered indefinitely by a wireless beam from the ground.

Reality in 2026

Wireless power exists at short ranges and in specialized applications. It did not become a practical replacement for conventional electricity distribution or the energy carried by commercial aircraft.

The International Energy Agency's aviation data still describe an aviation system that depends on substantial energy inputs and continued development of sustainable aviation fuels.^[41]

IATA's 2026 passenger-market reporting likewise describes an aviation industry built around conventional aircraft operations and fuel demand.^[42]

Prediction: Human Lifespans Would Become Extremely Long

A forecast associated with sociologist Hornell Hart in the mid-1920s claimed that babies born around the year 2000 could have something like 200 years of life ahead of them.

The same forecast suggested that centenarians would become much more common and remain vigorous rather than infirm.^[43]

Reality in 2026

Human longevity increased dramatically, but nowhere near the scale Hart imagined.

The CDC reported in January 2026 that U.S. life expectancy at birth reached 79.0 years in 2024.^[44]

The Census Bureau's current population materials also show a substantial and growing older population, but centenarians remain a small share of the population.^[45]

Prediction: Air Travel and the Motor Car Would Dominate Transportation

In 1926 centenary forecasts, David Dietz predicted that long-distance travel would be chiefly by air.

Cyril H. Bretherton predicted that the motor car would become the universal form of transportation.^[46]

Reality in 2026

Their broad transportation forecast came surprisingly close to the world that emerged.

IATA reported very high global passenger demand during 2026 and projected roughly 5.1 billion airline passengers for the year in its June industry outlook.^{[47] [48]}

Automobiles also remain central to everyday terrestrial transportation.

The forecast was not literally universal. People still depend on walking, rail, buses, bicycles, and other forms of transport.

So, What Does 2026 Actually Look Like?

Looking across these five eras, the reality of 2026 is not one simple future.

Some predictions were surprisingly close. Others missed their target completely.

Three patterns stand out in these examples.

1. Information Technology Was Often Easier to Predict

Predictions about information and communication were often closer to reality than predictions about large physical systems.

Nikola Tesla anticipated pocket-sized visual communication. Pew Research anticipated widespread mobile connectivity. Daniel Kokotajlo anticipated a major shift toward spending more computation on AI reasoning.

Physical systems were harder.

Commercial fusion remains unfinished. Ford's predicted 2021 Level 4 deployment did not happen. Commercial supersonic passenger service had not returned by September 2026. Tesla's vision of large-scale wireless power transmission did not replace conventional energy systems.

Physical technologies have to deal with materials, safety, infrastructure, regulation, economics, and real-world edge cases. Those constraints can make development much slower.

2. Human and Social Constraints Still Matter

Some forecasts were less accurate when they assumed technology would automatically change everyday behavior.

Auken's scenario imagined ownership being largely replaced by services. BT's forecasts pointed toward much more immersive digital environments. Kokotajlo imagined a much more thoroughly AI-shaped information environment.

Reality has been more mixed.

Services and subscriptions expanded. Mixed reality became more capable. AI-generated media became widespread.

But people did not simply abandon ownership, conventional screens, or existing social systems. Technology changed behavior without completely replacing older systems.

3. Short-Term Predictions Can Be Harder Than They Look

It is tempting to assume that a prediction made five years before an event should always be more accurate than one made a hundred years earlier.

The examples here suggest otherwise. They also resemble the idea commonly known as Amara's Law: people often overestimate the effect of a technology in the short term and underestimate its effect over longer periods.

Ford's 2016 forecast expected a production-ready Level 4 vehicle in only five years. Tesla's 1926 prediction of portable visual communication took roughly a century to mature.

Short-term forecasts can be shaped by project schedules, business incentives, and rapid extrapolation of current trends.

Long-range forecasts sometimes identify a technological direction without correctly predicting how long the journey will take.

Looking Back from the Future

To Nikola Tesla in 1926, 2026 was a wireless future in which geographic distance could largely disappear from human communication.

To Thomas O'Toole and Herman Kahn in 1976, 2026 was a distant technological checkpoint for energy, medicine, population, and economic development.

To automotive executives in 2016, 2026 was close enough for autonomous cars to seem nearly ready for mass deployment.

To AI researchers in 2021, 2026 was a plausible horizon for increasingly capable reasoning systems.

Today, we inhabit their destination.

The world of 2026 is an imperfect collage. We carry pocket computers that can transmit live video across oceans while still driving cars with steering wheels. We generate large amounts of electricity from solar technology while still depending on fossil fuels and nuclear power.

We also use increasingly capable artificial intelligence while living inside social, legal, and economic systems that change much more slowly than software.

Sources