Geoffrey Hinton put it plainly: we are at a point in history where something remarkable is happening, and nobody can say with confidence whether it will prove amazingly good or amazingly bad. Anyone who claims certainty is not being serious.

That uncertainty is not a reason for paralysis. It is a reason to look for better ways of understanding what is happening. In a recent session on navigating the AI, invention and IP landscape, Bill Fischer drew on established innovation theory to do exactly that.

His message was simple: while the future remains uncertain, the patterns of technological disruption are not entirely unfamiliar. The frameworks that have helped explain previous waves of innovation can also help us make sense of AI.

Change Comes in Waves

Fischer framed the current moment through the lens of S-curves. Technologies and competitive landscapes evolve in generations, progressing through periods of emergence, growth, maturity and eventual decline. Occasionally, that trajectory is interrupted by a rupture that creates a new curve altogether.

AI represents one of those ruptures.

Like many transformative technologies before it, AI has emerged from outside established industry trajectories. That helps explain why incumbent organisations often hesitate while new competitors appear from unexpected directions. Existing assumptions become less reliable, and new sources of advantage emerge.

Historical investment patterns illustrate the scale of such transitions. Major waves of capital expenditure accompanied the rise of railroads, electrification and the automobile. AI infrastructure investment is following a similar pattern, with some indicators suggesting the pace of investment is even faster.

Where Are We on the Curve?

According to Fischer, most AI applications (particularly “search”) are still in the early-growth phase of the industry’s S-curve. This stage is characterised by widespread uncertainty, a proliferation of competing approaches, the emergence of previously unknown challengers, hesitation among established leaders and a strong imperative to move quickly.

The current AI landscape reflects each of these characteristics. Large language models and AI assistants continue to multiply, while technology companies, start-ups and investors compete to establish positions in markets that have yet to stabilise. The eventual winners remain unclear because the market itself is still taking shape.

This uncertainty often creates the impression of chaos. From an innovation perspective, however, it is exactly what early-stage disruption should look like.

Designing for Adoption

One of Fischer’s most important arguments concerns adoption.

Drawing on Everett Rogers’ research into the diffusion of innovations, he noted that the success of a new technology depends not simply on what it can do, but on how quickly and broadly it is adopted.

Five characteristics consistently influence adoption: relative advantage, simplicity, compatibility with existing behaviours and workflows, trialability and observability.

Recent developments in AI provide useful examples. ChatGPT’s rapid growth was helped by its accessibility and ease of experimentation. Other platforms have gained traction by integrating AI capabilities into tools and workflows that users already understand.

The lesson is significant. Technical performance alone rarely determines market outcomes. Users adopt technologies when the value is visible, understandable and easy to experience.

For organisations developing AI-enabled products and services, adoption is not an afterthought. It is a central component of competitive advantage.

The Economics Behind the Race

Fischer also highlighted the economic realities underlying today’s AI competition.

The AI ecosystem can be viewed as a value chain with multiple layers. At the infrastructure level, cloud and platform providers are investing heavily while facing increasing pressure on margins as capabilities become more widely available. Foundation model developers continue to absorb enormous costs in pursuit of scale and market position. Application-layer businesses often benefit from attractive software economics, but many remain relatively small in absolute revenue terms.

Across all layers, the same principle applies: sustainable value creation still matters.

Growth narratives may attract attention and investment, but long-term success ultimately depends on business models that generate sufficient value for customers and sustainable returns for providers. The fundamentals of economics have not been suspended by AI.

What This Means for Human-AI Invention

Several conclusions emerge from Fischer’s analysis.

First, short-term confusion should not be mistaken for long-term weakness. Uncertainty is a normal feature of the early stages of technological disruption.

Second, differentiation matters. As markets evolve, organisations that create distinctive value are likely to be better positioned than those that simply replicate what others are doing.

Third, adoptability is itself a competitive variable. Technologies that fit naturally into human activities often spread faster than those that require people to fundamentally change how they work.

This may be particularly important when considering the future of invention.

Much of the public discussion around AI focuses on replacement: machines replacing tasks, jobs or even human creativity. Fischer’s framework suggests a different perspective. Technologies are adopted most rapidly when they make people more capable, more productive and more effective.

Viewed through that lens, the most significant opportunity may not be autonomous invention, but human-AI invention.

AI has the potential to make invention more democratic by expanding who can participate in the inventive process. It can accelerate invention once it pursues the identification of new opportunities and the generation of new solutions. And it can make invention more valuable by helping organisations connect technological possibilities with commercially relevant needs.

The organisations that succeed in the coming years may not be those that pursue automation for its own sake. They may be those that learn how to combine human insight, expertise and judgement with machine intelligence to create valuable inventions faster, more adoptable and at greater scale.

That possibility does not remove the uncertainty surrounding AI. But it does provide a useful way to interpret the current moment: not simply as a race to build better models, but as an opportunity to rethink how invention itself happens.

Bill Fischer, MIT presented “How to Navigate the AI, Invention and IP Landscape” in May 2026.

Watch the exclusive video here: