This is Where AI is a Loser
Updated: Oct 2
Discernment is a Land AI Will Never Inhabit
More and more, I am being asked: “How can one compete with AI, when it can already make things that look good, present lots of data, and are written pretty well?”
It’s a fair question.
AI tools can now generate words, sentences, images, layouts, and even whole interfaces and reports in seconds. They are fast, tireless, and increasingly polished.
But as AI increases the speed of generating drafts, options, and mountains of data, what matters more — and will matter increasingly - are human judgment skills: framing problems with informed intentionality; mapping constraints and limitations; evaluating what is actually “good,” valuable, useful, or enjoyable for users and contexts; identifying failure modes; making defensible trade offs, and knowing when the desired result has been successfully achieved.
These are all elements integral to how entities – and not just those in creative endeavors – interact with, and impact, our world.
What AI Is Actually Good At (And Why That’s Not Enough)
AI is very good at certain things:
Generating a massive array of options quickly
Remixing existing data, elements, and references
Producing “good‑enough” first drafts and variations
But AI cannot:
Understand the lived contexts where options will reside
Independently create and refine a new process or style
Know when something is, in fact, “finished” - or take responsibility when something falls short or fails, harms, or excludes.
AI is powerful at producing data and possibilities, but not at deciding the critical and interrelated factors of what is worth doing, for whom, and (especially) why.
Lessons from the World of Design: Problem Solving and Problem Finding
Years ago, a colleague of mine — now an in-demand product designer — told me something that was both simple and profound: designers are problem solvers. When I reminded him of this later, he said that he had thought more about that and had decided that – in order to solve problems - designers are also problem finders. They act with a sense of curiosity about the world, ever-refined abilities to effect change, and skilled intentionality to take that curiosity, apply abilities, resolve problems, and actually effect change. Aspects - all – that are foreign to AI.
I valued that insight because I realized it applied to any process requiring a solution or improvement. Consider, for example, how the humble stapler (or paperclip) answered a simple problem and revolutionized document management.
Think about it this way: AI can help to solve a problem - but who notices the need in the first place and frames the problem? Further, who decides which problem – of many - is worth working on? Who then sets the parameters of that problem – and then ultimately signals (knows) that the identified problem is “solved”?
To continue the design theme, in design schools, students learn to:
Ask “What is the real problem, need, or desire here?”
Reframe the ensuing questions in ways that reveal progressively refined issues (Associated mini-problems, if you will).
Establish criteria aimed at both the mini-problems (and their reduction) and towards solutions
Test, consider feedback, evaluate, re-test, re-evaluate, and validate the results.
Substitute any discipline – any discipline – for design here, and one realizes that this is not about being clever for the sake of cleverness, or art-y for artfulness.
It is about looking at the world with enough curiosity, care, and expertise to say: “There is the problem to be solved. Here is where I – or my team or discipline - can make a difference, [1] and this is what it will take to produce the desired results.” Results that are then evaluated against that “this” above – the established criteria.
Like a well-executed mosaic, each of the above-cited “this” and “that” elements are important in their own right.
But to create a successful mosaic – or result – requires more than merely recognizing the importance of the elements at hand.
 |  |  |
Unknown Mosaicist, Empress Theodora and her Court, San Vitale, Ravenna (Italy). 547. | ||
It requires seeing (understanding) each individual element, recognizing and establishing the correct relationships amongst those elements, and then progressing satisfactorily through those individually important elements and their relationships to a unitary – and verifiably satisfying – conclusion by exercising a skill that is seemingly simple – but is actually of immense complexity: Judgment.
Judgment
At its simplest, judgment can be reduced to the exercise of choice. As the mosaic example of related elements – decision points - makes clear, however, process judgments are rarely isolated moments of “this” versus “that.”
Consider the general applicability of the following concepts taken from the world of design. There, every design decision involves trade‑offs, such as: clarity versus density, familiarity versus innovation, cost versus craft, and more. Undergirding all of these is another critical aspect that is both a consideration and a desire: value.
Value, on the one hand, as: what is worth investing time, money, and effort in? And value, on the other hand (and often especially), as: what will genuinely make the resulting benefit better - more valuable - rather than just, say, flashier or more fashionable?
Such questions related to the above concepts, plus value, do not respond to formulas.
Nor do they respond well to the “this” or “that” matters of simple judgment.
They require much more. They require the ability to look at the interstices of relationships, possibilities, options, and solutions. It is – as designers put it - the ability to “see around corners” and to understand that critical inherent element that is – while not readily apparent – nonetheless discernable.
Discernment
Charles Spurgeon is credited with having written that discernment “is not a matter of telling the difference between right and wrong; rather it is telling the difference between right and almost right.” Or, for our purposes, between “good” and “good enough.”
More succinctly, perhaps, discernment is the ability to know when to color outside the lines – and why.
We often talk about discernment as if it were a mysterious talent or rare commodity. In practice, it is a set of learned habits and tested observations - plus the ability to understand what a given choice communicates or results in.
In an AI‑accelerated workflow, discernment is the most valuable tool of all. Let’s see why.
The Tools of Discernment
As noted, AI tools can generate hundreds of “solutions” in seconds. In a way – and here is where danger lies – it mimics discernment. Its answers – as imitations and approximations often do - look polished and refined. As students of the arts, design, literature and the humanities in general learn, mimicry is fundamental for learning, say, a style or technique.
It is insufficient, however, for knowing which style or technique (or outcome) is right for a given situation, audience, or culture. In short, outcomes divorced from meaning, context, or understanding– as in most AI output – are, well, meaningless.
Art, design, and the humanities help students of any discipline understand the nuances of meaning and outcomes: how elements of human activity and production have shaped societies, reflected power, framed messaging, influenced history, and have evolved across centuries, contexts, and continents. From art to advertising, science, politics, business and finance significance matters.
Perhaps the earliest surviving statement on the significance of such understanding in the professions comes from the first-century BCE Roman architect, Vitruvius:
“The architect should be equipped with knowledge of many branches of study and varied kinds of learning, for it is by his judgement that all work done by the other arts is put to the test.” [2]
Art, history, and cultural studies – hardly the province of AI as we know it today - are neither academic decoration nor mere “electives.” They are tools for discernment and judgment. They help students and practicing professionals recognize what certain choices can signal, why conventions exist, and – no matter the discipline – how outcomes have meaning and impact.
In design studies, it is a commonplace that design – as with outcomes - is never neutral. Choices carry consequences. Sometimes the choices lead to merely formal results. Clashing colors, for example. Other choices have more resounding consequences. Clashes of an entirely different variety.
That very human context is crucial when AI can generate “good enough” work – but without the meaningful context that refined discernment – and informed judgment - provides. Real discernment – genuine judgment - is never “good enough.”
As the Cheshire Cat in Alice in Wonderland put it: Any path will do if you don't know where you are going.
In short, someone has to make the call.
Discernment and judgment are quintessentially human skills. It is to this ability that AI can only aspire.
Until that time comes (if ever), AI will always fall short.
Tools will always change. New AI (or other) systems will appear and be replaced, and today’s “cutting edge” designs and interfaces will ultimately feel quaint and look curious on the shelves of technology.
But the ability to deal with ambiguity, frame problems, set intent, evaluate options, absorb feedback, and stand evidentially behind a decision that checks all the boxes— that is the durable and valuable skill.
Learning to Work With AI, Not Against It
So - to return from whence we came - the question to pose about AI is not how to compete with AI but: “What kind of professionals and leaders do we need in a world with AI?”
To try to compete with AI at speed or volume is a losing proposition.
The goal is to become – and educate - the kind of problem-solvers who can direct AI and who will know:
How to use it responsibly
When to use it
When to ignore it, and
How to integrate it into larger, human‑centered processes.
Taking cues from the world of design is not a bad place to start.
[1] It has become common to think about “making a difference” in the societal sense – and that applies here – but it can also apply to things like profitability, health, understanding, or even simple enjoyment, to name but a few examples.
[2] Vitruvius, The Ten Books of Architecture, M.H. Morgan, trans., 1960, I.I. - 1. Page 5 in Morgan.
Photo Credits
Designer with headphones: Nirmal Rajendharkumar on Unsplash
Toy robot: Emilipothèse on Unsplash
