The best, least expensive and most energy-efficient chip is the one between your ears.
It has been millions of years in the making. Can NVIDIA or any other technology company really outcompete evolution?
Let’s do a little thought experiment.
The human brain runs on roughly 20 watts of power. The energy in 1,500–2,000 kilocalories of daily food intake is equivalent to several hundred watts averaged over a day — although only a fraction of that energy is used by the brain.
A powerful AI GPU, meanwhile, can consume hundreds to thousands of watts while performing specialised computational tasks.
And yet, the human brain does something machines still struggle to replicate: general intelligence that adapts to an extraordinary range of situations, often with very little information.
We are nowhere near having a machine that can reliably match the full breadth of human intelligence.
Why?
Perhaps technology has two paths to catch up with evolution:
Time or energy.
Time: Keep improving algorithms, hardware and models, learning from nature’s extraordinary R&D programme — evolution.
Energy: Throw increasingly powerful chips, computing infrastructure and electricity at the problem.
Neither path comes free. Both have economic and environmental costs. And we still don’t know how far either will take us.
But perhaps we are overlooking something even more difficult to replicate.
Trust.
Intelligence is public. Trust is personal.
We have made enormous amounts of human intelligence available to the world. We express it through text, audio, video, products, services and code.
Technology consumes this accumulated knowledge, processes it using vast computing resources and generates outputs that can resemble human intelligence.
But intelligence is only part of what makes us human.
Trust is different.
Trust lives in our minds. It develops through experience, relationships, reputation and judgement. It can grow, weaken or disappear altogether.
We express some of our thoughts through language, but never everything. Much of our judgement remains difficult to articulate. We call it intuition, instinct or a gut feeling.
Perhaps these are imperfect labels for something we don’t fully understand.
Technology can process what we express. But can it truly understand what we leave unsaid?
And even if it could, would that make us trust it?
Can trust be artificial?
We are naturally predisposed to trust. It has helped humans cooperate, build communities and accomplish things that no individual could achieve alone.
Trust is an undervalued human superpower.
Intelligence can be artificial. But there is something odd about the phrase artificial trust.
It sounds almost untrustworthy.
Of course, technology can help us verify information, enforce rules and reduce the need to trust individual actors. Blockchain, for example, attempts to enable certain transactions without relying on a central intermediary.
But replacing one form of trust often means relying on another: the technology, its designers, its governance or the institutions supporting it.
Trust doesn’t necessarily disappear. It changes address.
The economics of trust
Chris Davis of Davis Advisors inspired me to think about trust in economic terms.
His argument highlights an intriguing question: What does it cost to replace trust with technology?
Consider a business where people trust one another.
Decisions can be made faster. Fewer resources are spent checking every action, negotiating every detail or protecting against every possible failure.
Now imagine removing that trust.
We need more controls, more verification, more documentation, more monitoring and more enforcement. Some of these are essential, of course. But they all consume time, money and resources.
Technology can reduce some of these costs. Yet building and maintaining technological systems also requires infrastructure, computing power and energy.
As Davis suggests, trust is not merely a moral virtue. It is an economic asset.
You don’t have to believe in the power of goodness to recognise its value. You can be entirely pragmatic and still conclude that trust is an extraordinary source of efficiency, particularly in business.
A team that trusts one another can often achieve more with less.
The final frontier
Trust is a prerequisite for high performance and harmony in teams, families and societies.
When trust breaks down, we compensate with rules, checks, controls and technology. Sometimes we must. But these substitutes can become expensive, and they rarely recreate everything that trust makes possible.
AI may become more intelligent. Machines may become more capable. Computing may become cheaper and more energy-efficient.
But the bigger question is not simply whether technology can imitate human intelligence.
Can we build a world where technology makes trust stronger rather than making trust unnecessary?
Replacing human trust with technology may be the last frontier.
We are not there yet. Perhaps we never will be.
And even if we could cross that frontier, should we?
Read all my “Notes to Self” at view all blogs.