Despite what the AI hustle bros would have you believe, tech is a perfect answer for only a vanishingly small number of the world’s problems. Like writing software, for example, which is entirely logical and sequential, and you can train a machine to write it.
For almost everything else, tech is, at most, a supporting cast member in a production that’s 90% or more “not tech”.
If you don’t believe me, consider this: recently Bill Gates said that AI would replace most teachers in the next 10 years.
Now Bill Gates is vastly more successful than I’ll ever be, so I’m not personalising this or being disrespectful. It’s just a timely illustration of how so many tech folk (including fabulously wealthy ones) completely misunderstand reality outside their entirely logical binary world, populated entirely by 1s and 0s.
However the reason I’m writing this article is that the way tech folk think about the world is an easy trap to fall into. And these are traps that, from time to time, non-tech businesses fall into as well.
So, to help you avoid those traps, here are some of the ways tech bros and tech gals fail to appreciate just how little of the world they are in a position to impact, whether that’s inside the classroom or anywhere else.
The world is irrational
Whether we like it or not, the world is a highly irrational place. Tech is an almost entirely logical place.
This is one of the main reasons tech folk get it so badly wrong.
Rather than accepting the world is irrational they (ironically) irrationally assume the world is entirely logical and design all their tech solutions on the assumption that this is an accurate representation of reality, even though purely logic-based situations probably represent less than 10% of reality for most of us.
The politicians we elect, the car we drive, the food we buy, the places we live, the people we fall in love with – none of those are logical decisions.
Sure, sometimes we post-rationalise a logical excuse for our decisions. But just because we can articulate a logical-sounding post-hoc justification for it, that doesn’t mean the original decision was logical.
It just means we’ve learned to cover our tracks in a world that thinks rationality is a strength, rather than a weakness.
That’s why we don’t boast that our new BMW makes us look super-cool (well, most of us don’t – a few psychopaths do…). Rather we talk about the fuel economy, or the 0-60 speed, or the excellent residual values.
The truth is, we thought the Beemer would make us look cool. That’s why we bought it. But we post-justify our decision with a superficially logical excuse for doing what we (irrationally) wanted to do in the first place.
In case you think I’m being a little harsh here, there are cars every bit as good as BMWs. They’re just as economical at the petrol pump. They get to 60 in more or less the same time. The lower purchase costs can make their “whole life” financials very similar to a fancy BMW’s.
Yet, BMW can sell their cars for twice what some other, logically comparable models sell for.
The decision to buy a BMW – and almost anything else you buy – is almost entirely irrational.
So don’t make the mistake that tech bros and tech gals make, and assume that the world is a logical place. Perhaps less than 10% of it is – and, because that’s how human beings are wired, that’s unlikely to change in the next several millennia.
If you make business decisions assuming everyone is logical, you’ll be wrong an astonishing amount of the time.
Consider this: if you were entirely logical, every cupboard in your kitchen would be full of Tesco Value products. And, if you’re reading this, I’m prepared to bet that there are zero to not-very-many Tesco Value products in your kitchen. Everything else in there was an entirely irrational purchase, if you followed the same logic as tech folk think you ought to.
So if that’s true for your kitchen, how true do you think that might be in every other purchasing decision you, and your customers, make?
Tech folk think that pretending the world is entirely logical is the best approach. Given that reality suggests that assumption is almost entirely wrong, why would you make decisions in your business which assumed that a state of affairs which might exist, at most, 10% of the time should govern your decisions 100% of the time?
Teaching vs learning
I spent over a decade in the education sector, and one thing that always struck me was that there are two words we civilians use more or less interchangeably, but which are entirely different in practice.
In education, teaching means someone standing at the front of the room on “broadcast mode”. They read their slides while 30 pupils stare out the window, scribble in their books, or pass notes to their friends.
At one level, a KPI has been achieved. Year 7 has indeed experienced 45 minutes of geography teaching.
But if, during that 45 minutes, nobody has paid attention, and couldn’t recall a single thing the teacher spoke about, it’s all been a bit of a waste of time.
Learning, on the other hand, is what the experience looks like from a student’s perspective. Did they learn something they didn’t know before? If they did, the goal of the education system has been satisfied – Year 7 are now more educated than they were before the class started.
Tech bros and tech gals, Bill Gates included, it seems, confuse teaching with learning and struggle to conceive that there might be a difference between those two concepts.
That’s because it’s really easy to design a metric for teaching – did Mrs Smith turn up at 10am and talk for 45 minutes about geography?
And it’s really hard to design a metric for learning. How do you really know what went into a student’s head while they sat in Mrs Smith’s classroom?
There’s an attempt to do a vague post-hoc assessment by making students sit A levels and the like. But if the exam question is on a topic that Mrs Smith covered in April, and you fail it, does that mean you learned nothing from Mrs Smith in March or May?
Of course it doesn’t. Even overlooking its technical merits as an assessment approach, an exam is an extremely poor basis for assessing what a student has learned because it’s impractical to examine every part of the whole year’s geography syllabus in a 2-hour exam.
You’re just throwing darts at a dartboard and hoping a few of them land.
And the corollary is true too. If I was only awake in Mrs Smith’s class during April, but that happens to be when she covered the topic which we get examined on, I look like a genius even though I’ve learned a tiny fraction of what the syllabus says I should have learned.
(Full disclosure: I’m pretty sure this is how I got a wholly unexpected B in my Higher Chemistry, to the open-jawed astonishment of both my father and my chemistry teacher. That’s a class I was almost certain to fail.)
I can’t leave this section without noting that it’s not only tech folk who have problems with this. Ofsted, the government body which is supposed to assess education standard can’t seem to understand the difference between eminently KPI-able “teaching” and much harder to assess “learning” either.
Of the two, learning is what really matters. Teaching is just the pre-packaged way of getting about 10% of the way there.
AI will almost certainly underperform human teachers, except the truly dreadful ones, because the huge difference an enthusiastic, committed teacher can make to any subject can’t be replicated by non-humans.
I won the geography prize every year in senior school because I had a passionate, committed teacher who made the subject interesting…unlike the chemistry teacher did for their subject, for example.
The teaching equivalent of a bank’s AI chatbot reciting facts about geography would not have made me as enthusiastic about geography as my human teacher did.
Knowledge, not facts
To compound the problem of tech bros and tech gals confusing their perspective (eg teaching) with the customer’s perspective (eg learning), the next mistake they make is confusing knowledge with facts.
We want students, young and old, to have knowledge, not just to be able to regurgitate facts.
In the real world, outside the AI bros’ laboratory conditions, facts are almost trivial commodities. What really counts is knowledge.
In a business setting, a fact might be a KPI. It’s a number or a score of some sort, and the result of a mathematical calculation.
But what you’re really interested in as a business leader is why that number is what it is – whether that’s good, bad, or indifferent.
A fact, or a KPI, is knowing that the factory output was 50% of target last week.
Knowledge is what really matters, though.
Your response to that fact will be entirely different depending on whether it’s because a key piece of machinery broke and it took 48 hours to get a new part because your finance department hadn’t paid your key suppliers for months…or because the King was hosting a state visit to your factory and it had to be shut for 48 hours to enable all the security checks to take place…or your entire staff had been on an all-weekend bender the previous weekend and were too inebriated to operate the machinery safely for a couple of days after they got back.
Facts have a role of course. And in some areas of business more than others – if you ship products to a 99.9% purity level, you’d better have some factual way of assessing whether or not that standard has been reached.
But I’ve very rarely come across a business whose main problem is a lack of facts. Nearly always, businesses which commit monstrous acts of self-harm do so because of a lack of knowledge about what those facts mean. (Or a lack of accurate knowledge, at least – they nearly always have reams of eloquent explanations for their perilous predicament which are almost entirely wrong.)
I’m quite prepared to believe that AI can deliver facts to an 80% accuracy over the next decade or so (up from, I’m guessing, about 40% now).
But AI teaching you that World War Two started in 1939 – even if that’s a fact you happen to remember – means almost nothing.
What really matters is how we got to the point of a global war breaking out in the first place, and how the world dealt with its aftermath, which heralded a new world order, a state of affairs which still has a powerful hold over the world as it is today.
They are the aspects we need to learn from, and AI is unlikely to go a good job of conveying that any time soon.
It might well be able to tell you that a key fact in the build-up to WW2 was Neville Chamberlain’s visit to Munich in 1938. But that visit was so famously unsuccessful for a whole range of reasons which are entirely irrational.
And, as we saw above, tech can’t cope with irrationality – it can only operate in conditions of perfect rationality.
In your business, you want people to have knowledge, not to just spout facts – no matter how beautifully they present their performance charts (aka “facts”) in Power BI.
Facts are commodities with very little value. Knowledge is everything.
The 7% rule
To be clear, I don’t believe precise statistics when it comes to fundamentally unmeasurable situations. However I think there’s a considerable amount of truth underlying this message, even if I don’t think the % splits are necessarily always accurate.
Which, by the way, also illustrates why facts and knowledge are independent from one another. I can take a broad message from a statement which is helpful at some level whether or not it’s actually factually accurate to seven decimal places.
What I’m talking about here is the statement much loved by business coaches and leadership trainers that only 7% of your communication with another human is made up of the words you speak.
The rest is your tone of voice, your physicality (eg do you have a smile or a frown on your face as you say the words) and a range of other factors.
That 7% doesn’t need to be factually accurate for us to understand that only a tiny part of our communication is dependent on the actual words we use. The facts, if you like.
If your life partner asks, “what would you like for dinner?” the actual words spoken are doing very little of the communicating here.
The real message could be anywhere on a spectrum from “aw, you look really tired tonight so I’ll make dinner even though it’s your turn” to “I’m beside myself with fury that you forgot my mum is coming around for tea and you were supposed to have everything ready for serving up by now!”.
The exact same words. Two entirely different meanings.
For tech folk, words are all they work with. Programming tech is a series of logical words, instructions, and if/then decision points.
So whether the number is 7%, 12% or something else, we can probably agree that only a minority of our communication is the result of the words we speak.
But that’s all tech folk can work with. The surface vocabulary, and the formal dictionary definitions, not the meaning behind the words.
It’s why tech folk make the fundamental error of assuming our post-hoc rationalisations for our purchasing decisions are the real underlying reasons we bought something.
Because that’s the bit in words. The bit of your brain which is imagining how cool you’ll look to all the ladies you fancy is invisible to any form of tech.
This is also why your business can issue 80-page instruction manuals which explain everything people are supposed to do and you’ll still find people not doing the things the instruction manual says they should.
It’s not that they’re incapable of understanding the words. It’s because that’s only 7% (or whatever) of the total amount of communication required for them to fully understand.
I don’t think this is a very difficult concept to get your head around, but apparently it is for tech bros and tech gals.
However brilliant it might become, the maximum contribution tech can make to conveying understanding is just 7% of what is required, because that’s the maximum amount which can be conveyed by words alone.
All the other parts of the communication process require entirely human characteristics like empathy, understanding, appreciation for another person’s point of view, the wider cultural and societal context within which this communication is taking place, whether the person you’re trying to communicate with has eaten in the last 48 hours, and a whole host of other factors that tech does not know and never will know.
Sure, tech can do “pretend empathy” but that has zero impact with anyone. If it had a remotely positive impact, you’d spend a lot more money than you do on products promoted in all those emails you receive addressed to “Dear {Firstname}”.
Maybe in a thousand years, tech will be able to convince you it’s being empathetic by stimulating electrodes planted in your head. But until then nobody is fooled by the sort of formulaic faux empathy you get from tech solutions.
And, to return to Mr Gates’ assertion, that’s why teachers will not be replaced in any meaningful sense by AI. Because even great AI maxes out at 7% of the communications spectrum.
A great teacher can easily 10x that performance and AI has no way of catching up because it’s incapable of exceeding 7%…no matter how good the tech becomes.
Reality trumps theory every time
Tech folk, and perhaps Mr Gates, dream of theoretical worlds where everything is perfectly organised and logical. The world isn’t remotely like that, of course, but that’s what tech bros and tech gals think is going on.
What they forget is that you can theorise all day long – and I get it, this can be an intellectually stimulating process if your brain is wired that way – but reality always trumps theory.
All the business world’s biggest mistakes have been made because organisations think their theories represent reality, when they don’t.
Coca-Cola conducted exhaustive tests before launching New Coke – theoretically a “better” product.
Except nobody wanted to buy it. Coca-Cola was forced into a humiliating climbdown almost immediately because the reality of the market trumped the theory in their labs and their boardroom.
The Ford Edsel – one of the world’s biggest car manufacturers worked incredibly hard to develop a product to a rationally-optimal solution, only to find that nobody wanted to buy the car that came out of the end of that process.
In theory, securitised sub-prime mortgages are a great idea. The reality is that the way they were sourced, and later packaged up, created the conditions which nearly destroyed the world’s financial system.
I could go on, but you get the idea.
There’s an old army saying that “no plan ever survives contact with the enemy”.
In a business setting, the reality of the market counts for a lot more than the theory of what ought to happen, as Coca-Cola, Ford and plenty of others can very fully testify to.
The problem is that AI, and tech more generally, can’t deal with reality all that well.
Reality is messy, illogical, and powered by human thought processes which are undetectable to our robot overlords.
Tech has no way of even knowing what reality is, much less understand it, process it, and adapt to it.
Again, in a superficial way, tech claims to, but that’s mostly garbage as anyone who ever bought context-sensitive advertising knows.
When adverts for package holidays appear next to news stories about plane crashes because the requisite keywords in the news article triggered the ad placement, you know that particular tech solution is a busted flush.
In an educational sense, teachers have to factor in student’s health, well-being, home circumstances, problems with drugs or alcohol in the family, and a litany of other factors if they want to communicate with every student to the maximum extent possible.
Tech solutions wouldn’t even know where to start with all that. They have no way of collecting and processing that information. And likely never will.
Teachers are here to stay
So, with respect to the fabulously wealthy Mr Gates, the chances of AI making most teachers redundant any time soon is pretty low.
Even though I’m sure the tech will improve, and the more evangelical wing of the tech industry will shout loudly about how their secret alchemy will change the world for the better, the reality is that no matter how good tech gets, we will only move from tech handling perhaps 2% of a communications process at the moment up to handling 5% of it, out of the potential maximum 7% of communication that is based on facts alone.
While that’s an impressive improvement on the current performance, delivering 5% of the communication through tech still leaves 95% of a communications process “un-tech-able”, apart from a few tricks like personalising salutations on sales emails to make you think the tech cares about you, which fool no-one.
Sure, if you think facts are the only element of education that matters, and that people operate like software programmes, even though all the available evidence is that they don’t, then tech might play a role.
But the reality is we could spend £billions on AI assisted teaching solutions, at great expense to taxpayers, and barely move the needle when it comes to building a well-informed workforce who understands what they do well enough to make a difference to the UK economy and the lives of everyone who lives here.
Of course, tech has a role. But if you think “AI everything” is a good idea, just remember this is like asking your Finance Department to run every aspect of your business – from strategy, to production, supply chain management, HR, sales, marketing, and more.
If you think that’s a good idea, implement as much tech as you like. At the very least that’s likely to hasten the inevitable and unpleasant end, so you can get on with doing something more productive with your life.
But if you think that would be a crazy idea, then think twice before you get seduced by the “AI is the future” crowd, no matter how wealthy and revered the person telling you that might be.
For that idea to work, you need to believe there’s no “human” in “humanity”.
And I, for one, don’t want to be just an “ity”.