There was a time when one of the biggest obstacles to making a good decision was simply getting enough information.
You needed access to information and the expertise to locate and make sense of it.
Sometimes you even needed what we used to call an “expert” – someone who had spent years mastering a certain skill.
That constraint is disappearing remarkably quickly.
Today, you can give AI details of your income, assets, expenditure, age, retirement goals and risk tolerance and ask it to help you think through your financial future.
Within seconds, it can produce projections, compare strategies, identify considerations you may have missed and construct something that looks remarkably like a financial plan.
And this isn’t some distant hypothetical.
Vanguard now describes AI as capable of analysing enormous datasets, optimising portfolios and updating financial plans in real time. The analytical work that once consumed enormous amounts of professional time is becoming increasingly automated.
That sounds like extraordinarily good news.
And in many ways, it is.
But it also exposes something we’ve misunderstood about difficult decisions for a very long time.
We Thought We Had an Information Problem
Think about how we usually respond when we’re uncertain about an important decision.
We research.
We read another article.
We watch another video.
We ask another expert.
We compare another option.
The assumption underneath all of this is perfectly reasonable:-
If I can just get enough information, the right answer will eventually become obvious.
For most of human history, that wasn’t an entirely bad assumption.
Information was scarce.
Expertise was expensive.
Analysis took time.
AI has changed all three.
Which creates an interesting test.
If lack of information were the primary reason intelligent people struggled to make good decisions, AI should be rapidly eliminating the problem.
Instead, many of us now have something quite different.
More answers than we know what to do with.
AI Has Made Answers Cheap
Ask AI whether you should put more money into equities.
You’ll get a perfectly plausible analysis.
Ask whether property should form part of your retirement strategy.
Another plausible analysis.
Ask about bonds, private markets, annuities, dividend investing, entrepreneurship or maintaining some form of earned income beyond traditional retirement.
Again, you can receive a sophisticated answer in seconds.
Then change one assumption.
Or ask the question differently.
Or introduce another objective.
Suddenly, another course of action may look equally reasonable.
This isn’t necessarily a failure of AI.
It’s the nature of the decision.
Because many important decisions don’t have a single objectively correct answer waiting to be discovered.
They involve trade-offs.
Risk.
Incomplete information.
Competing objectives.
And a future nobody can know with certainty.
AI can help us analyse those things.
It cannot make the uncertainty disappear.
More Information Doesn’t Necessarily Create More Certainty
This is where something counterintuitive happens.
Once we have access to almost unlimited analysis, we don’t necessarily make decisions more easily.
Sometimes decisions (in particular decisive ones) become much more difficult.
Every answer creates another question.
Every strategy has a counterargument.
Every opportunity has a risk.
And every confident prediction can be matched by someone—or increasingly, something—capable of constructing an equally convincing argument for the opposite conclusion.
So we keep looking.
Not because we lack information anymore.
Because we’re hoping the next piece of information will give us something information cannot provide…
certainty.
That’s a very different problem.
The Bottleneck Has Moved
For decades, knowledge was enormously valuable because acquiring it was difficult.
AI is rapidly changing the economics of knowledge.
That doesn’t make knowledge worthless. Far from it.
But it does change where the constraint lies.
When information becomes abundant, the scarce resource becomes the ability to decide what matters.
To distinguish evidence from prediction.
To recognise which assumptions should apply.
To understand the downside if you’re wrong.
To weigh competing objectives.
And eventually, to make a decision without pretending that things are no longer uncertain.
In other words:
While AI has made information cheaper. It has made good judgement more valuable.
Vanguard makes a related distinction in its work on AI and financial advice.
It argues that data gathering, portfolio construction and financial-planning simulations are increasingly susceptible to automation, while interpretation, behavioural guidance and understanding what actually matters to someone remain much more dependent on human judgement.
In my view, the implications go well beyond financial advice.
The Same Problem Appears Everywhere
Consider almost any important decision.
Should I leave my job?
Should we buy this business?
Should I invest in this opportunity?
Should I retire now or work another five years?
Should I start something of my own?
Should we enter this market?
AI can make you dramatically better informed before making any of those decisions.
That’s valuable.
But eventually you encounter the same boundary.
The available evidence ends.
The future begins.
And someone still has to decide.
Intelligence and Judgement Are Not the Same Thing
We tend to associate good decisions with intelligence.
Understandably so.
But intelligence gives us an unusual problem.
The better we are at analysis, the better we can become at constructing arguments for multiple possibilities.
We can see more variables.
More risks.
More scenarios.
More second-order consequences.
AI magnifies that capability enormously.
Give it one course of action and ask for the strongest argument against it.
Then ask it to rebut that argument.
Then ask for three alternative strategies.
Then ask what you’ve overlooked.
You could continue almost indefinitely.
At some point, more analysis stops reducing uncertainty.
It merely describes the uncertainty in greater detail.
That’s when judgement begins.
Good Decisions Have Never Required Certainty
This distinction matters enormously in finance because financial decisions often involve consequences that won’t become clear for years.
A good outcome doesn’t necessarily prove that the decision was good.
A bad outcome doesn’t necessarily prove that it was bad.
You can make a sensible decision based on the information available today and still experience an unfavourable result.
You can also make a reckless decision and get lucky.
That’s why I think one of the most useful disciplines is to judge a decision based on what was knowable at the time you made it, rather than rewriting the quality of the decision after you discover the outcome.
AI can increase what the span of what is knowable.
That is a tremendous advantage.
But it cannot turn the unknowable future into the knowable.
And we shouldn’t ask it to.
The Skill We May Need Most in the AI Age
Much of the discussion about artificial intelligence focuses on which human skills AI will replace.
I suspect an equally interesting question is which skills become more valuable because AI exists.
Judgement may be one of them.
Not intuition masquerading as judgement.
Not confidence.
Not prediction.
The disciplined ability to gather the available evidence, understand the assumptions, assess the downside, recognise what remains unknowable and still make a considered decision.
Because the strange consequence of having nearly unlimited access to answers is that answers themselves become less valuable.
The ability to know what to do with them becomes more valuable.
AI can give us more information than any generation in history has possessed.
It can help us model possibilities our grandparents couldn’t have calculated in a lifetime.
It can challenge our assumptions and expose things we’ve overlooked.
We should use it.
But perhaps the greatest mistake we could make is assuming that better answers eliminate the need to become better decision-makers.
They don’t.
They make it more important.
Because AI may eventually make information almost unlimited.
Our ability to act intelligently when certainty is unavailable will remain scarce.