Remember when Google Glass was going to revolutionise how we interact with technology? When 3D TVs were the inevitable future of home entertainment? When blockchain was going to decentralise everything? When the metaverse was going to replace the real world?

Tech predictions age like milk. Within a few years, most bold claims about the future look embarrassingly wrong. Yet we keep making them, and people keep taking them seriously.

This isn’t just harmless speculation. Bad tech predictions shape investment decisions, policy choices, and careers. Understanding why predictions fail—and why we keep making them anyway—matters.

The Fundamental Problem

Predicting technology is hard because it’s not just about technology. It’s about human behaviour, market dynamics, regulatory responses, unforeseeable events, and complex interactions between all of these.

A technology can be brilliant but fail because the timing’s wrong. Or because a competing standard wins. Or because regulation blocks it. Or because users don’t want what technologists think they should want. Or because it creates problems nobody anticipated that kill adoption.

Tech predictions typically focus on technical feasibility. “This technology can do X, therefore people will use it for X.” But feasibility doesn’t determine adoption. Most failed tech predictions were technically sound. The technology worked. It just didn’t matter.

The Hype Cycle Is Real

Gartner’s hype cycle model describes how new technologies go through predictable stages. Initial excitement, overinflated expectations, disillusionment when reality doesn’t match hype, gradual productive adoption.

Most tech predictions are made during the peak of inflated expectations. That’s when the technology is new, exciting, and easy to imagine transforming everything. That’s also when we’re least capable of accurate prediction.

Once the technology enters the trough of disillusionment, predictions shift to the opposite extreme. “This will never work, it was all hype.” Then, quietly, the technology finds actual useful applications and gets adopted for boring practical reasons nobody predicted.

Smartphones went through this. Early predictions said they’d replace everything—wallets, cameras, computers, everything. Then initial versions were clunky and disappointing. Then, gradually, they actually did replace a lot of things, but not in the ways initially predicted.

The pattern repeats with every new technology. AI, VR, blockchain, quantum computing—all following the same hype cycle with predictions that wildly overshoot during excitement and overcorrect during disillusionment.

Incentives Reward Boldness

Nobody gets attention for predicting incremental change. “Technology will continue improving gradually” is true but boring. “This technology will completely transform society” is exciting and gets clicks.

Tech commentators are incentivised to make bold predictions because bold predictions generate attention. Even if you’re wrong, being wrong in interesting ways is better for your career than being right in boring ways.

This creates selection bias. The predictions you hear are the most dramatic ones, not the most likely ones. The prediction marketplace rewards confidence and extremism over accuracy and nuance.

And because wrong predictions are quickly forgotten while right predictions are heavily promoted, successful predictors often have terrible overall track records that nobody remembers because we only recall their hits.

The Visionary Fallacy

We lionise people who predicted major technology shifts. Steve Jobs predicting mobile computing. Marc Andreessen talking about software eating the world. These become legendary forecasts that cement reputations.

But we ignore all the predictions these same people got wrong. Jobs thought the iPod would never need more than 5GB. Andreessen was wrong about numerous specific companies and technologies. Their legendary status comes from cherry-picking successes and ignoring failures.

This creates a “visionary” class of predictors whose track records don’t actually support treating their predictions as particularly reliable. But because they were right about some big things, we assume they have special insight into the future.

They don’t. They’re just people who make lots of predictions, some of which inevitably hit. The apparent insight is mostly survivorship bias.

Technical vs. Social Prediction

Predicting what’s technically possible is very different from predicting what people will actually do with technology.

Engineers are generally pretty good at predicting technical capabilities. Moore’s Law held for decades because semiconductor physics is predictable. We can reasonably forecast computing power, storage capacity, bandwidth improvements.

But social adoption is chaotic. Will people want this? Will they trust it? Will it fit into their lives in useful ways? These questions involve psychology, culture, economics, and sociology. They’re much harder than technical questions.

Most failed tech predictions get the technology right but the social response wrong. Google Glass technically worked. People just didn’t want to wear computers on their faces. 3D TV technology was fine. People didn’t want to wear glasses to watch television at home.

Technical possibility doesn’t create social demand. But tech predictions often assume it does.

The Disruption Obsession

Tech culture is obsessed with disruption. Every new technology is framed as disrupting existing industries, replacing old ways of doing things, making established companies obsolete.

This happens sometimes. But usually technology integrates into existing systems rather than replacing them. Incumbent companies adapt. Regulations shape adoption. New technologies coexist with old ones longer than predicted.

The disruption narrative is appealing because it’s dramatic. It implies that current winners will become losers and new players will triumph. It’s the tech version of revolutionary politics—total transformation rather than gradual evolution.

But gradual evolution is usually what happens. Technologies like bespoke AI development show how new capabilities tend to augment existing business processes rather than completely replacing them. The boring reality is integration, not revolution. But integration doesn’t generate exciting predictions.

What Actually Works

Some approaches to tech prediction are better than others.

Trend extrapolation works for certain things. If something has been improving exponentially for decades, betting on continued improvement is reasonable. This works for processor speed, storage, bandwidth—things governed by physics and economics with long track records.

Analogies to previous technology adoptions help. Smartphones were like previous computing platforms but mobile. Streaming was like previous media distribution but internet-based. New technologies often follow patterns set by earlier ones.

Watching actual early adopters matters more than listening to visionaries. If a technology is genuinely useful, you’ll see people solving real problems with it early on. If early adoption is mostly marketing demonstrations and tech demos, it’s probably not going to scale.

Being specific about mechanisms helps. “AI will transform healthcare” is too vague to be useful. “AI-assisted diagnosis will reduce errors in radiology by helping doctors spot patterns in scans” is specific enough to evaluate. Good predictions include plausible mechanisms.

Why We Keep Doing This

If tech predictions are usually wrong, why do we keep making them?

Because they’re useful despite being inaccurate. Predictions help organisations plan, even if the specific predictions are wrong. Thinking about possible futures helps prepare for uncertainty. The process of making predictions forces consideration of trends and possibilities that might otherwise be ignored.

Predictions also serve cultural functions. They express hopes and anxieties about technology. They let us collectively imagine different futures. They’re stories we tell about where we’re going, even if we don’t actually end up there.

And sometimes predictions are self-fulfilling. Enough people believing “mobile is the future” made mobile the future by directing investment, attention, and effort toward mobile technologies. The prediction shaped reality by influencing what people built.

The Right Way to Use Predictions

Treat tech predictions as scenarios, not forecasts. They describe possible futures, not inevitable ones. Use them to think through implications and prepare for possibilities, not to bet everything on specific outcomes.

Pay attention to who’s making predictions and why. Venture capitalists predict things that would make their investments valuable. Tech companies predict things that would increase demand for their products. Consultants predict things that would create demand for their services.

This doesn’t mean their predictions are wrong, but it means you should consider incentives.

Focus on underlying trends rather than specific technologies. “People want convenient access to information” is a more useful prediction than “Google Glass will succeed.” The trend is real even if the specific technology fails.

Be skeptical of predictions that require human behaviour to change dramatically. Technologies that fit into existing behaviour patterns succeed more often than technologies that require people to change.

The Humility We Need

Tech predictions would be better if predictors were more humble. Admit uncertainty. Give probabilities instead of certainties. Acknowledge that you might be wrong.

But humility doesn’t generate attention, so we’re stuck with overconfident predictions that are usually wrong but occasionally spectacularly right.

The best you can do is consume tech predictions with awareness that they’re mostly entertainment, occasionally insight, never certainty. Enjoy the speculation while remembering that the future will surprise everyone.

Because the thing about the future is that it’s actually unknowable. All we can do is make informed guesses and prepare for multiple possibilities.

Tech predictions will keep being wrong. We’ll keep making them anyway. And that’s fine, as long as we remember they’re guesses, not prophecies.

The future will arrive whether we predict it accurately or not. Might as well have fun being wrong about it together.