To be investable, ed tech has to be both academically credible and honest about its aims and limitations. Joe Francis, the former Eton master who founded Brillder, sorts pseudoscience from substance.

Warren Buffet famously advised: “Never invest in a business you cannot understand”. There is some irony in the fact that while education is supposed to be about helping you understand stuff, the education technology business often shrouds itself in the gobbledygook of acronyms and pseudoscientific jargon.

Of course, lots of respectable professions like medicine and law use sophisticated, specialised terminology. But even among medics, just occasionally, a quack is exposed – someone who may even sound like he knows what he’s talking about. In ed tech, I have come to wonder whether endemic quackery contaminates the way we try to sell our product – both to consumers and investors.

AI is the biggest tech buzzword right now. But it’s becoming fairly clear that a larger number of ed techs claim to be using artificial intelligence than actually are. I recently heard a chief executive explain how AI created personalised learning on her platform by modifying the presentation of material in response to learner performance. What I think she was actually describing were adaptive pathway templates, a technology which has nothing to do with AI and has been around more or less since the invention of coding. But this did not inhibit the success her company has had both marketing her platform and raising money from investors, some of whom may imagine AI is spontaneously creating questions and explanations. It’s not. In fact, content – at least content that is any good – is always authored by human beings.

Is it conceivable that both ed tech customers and ed tech investors are reassured by the jargon of innovation and hi-tech pedagogy? Do they wonder whether ‘asynchronous’ and ‘personalised’ learning are the same or different things, or worry about the difference between v-labs and VLEs? Perhaps they just feel in safe hands when they hear such terms.

Angel investors are generally less interested in the product than the business (frustrating for founders trying to launch a game-changing product). If user growth is encouraging, if the idea is demonstrably monetisable, these are the most important things. But what seems to tip investors towards or away from a deal is the sense that something is extra clever about a platform, that there’s a cutting ‘sciencey’ edge and perhaps some valuable IP.

I think education technologists play up to this a little bit. Possibly quite a lot.

Many platforms proudly tell you, for example, that they use neuroscience to optimise learning. Neuroscience, or neural science, is really about the nervous system, not just the neural circuits of the brain. What they may mean is that their approach is consistent with findings in cognitive science.

However, lots of other industries could make similar claims, including gaming and publishing. But in publishing you never hear guff like “our books organise information into sentences, paragraphs and chapters, using neuroscience to aggregate micro-units until the reader achieves macro-cognition”. And in the gaming industry, of course, they don’t want to boast that your neurotransmitters are being constantly saturated in dopamine – after all, even cocaine uses neuroscience.

The problem is that, as a whole, ed tech is insecure. There are two main reasons for this. The first is that while tech is quite good at delivering and testing knowledge, it struggles to serve the more prestigious aims of education which address open questions and uncertainties. The second is that whereas in most industries, such as banking, retail or travel, technology has improved productivity by replacing human beings, there is no evidence that ed tech is (as so many of its entrepreneurs love to claim) ‘transforming’ education, either in terms of productivity or effectiveness. In the UK, where expensive private schools are actually losing market share, private tutoring is probably booming more than ed tech (indeed, several successful ed tech platforms are essentially interfaces for private tutors). For many learners, buying the time of an intelligent human being is better value than software.

Perhaps even the claim that ed tech can test knowledge effectively needs scrutiny. If a chemistry student is asked: “What is the melting point of cadmium?”, the answer is only as good as the ones which are permitted. If, for example, the input only recognises a superscript circle to denote degrees (that is, 321.1𐩑 C) then not only might the word ‘degrees’ not be accepted, but the use of a lower case ‘o’ or numerical zero might invalidate a correct answer. And if you happen to write “six hundred and ten degrees Fahrenheit”, computer definitely says no.

To avoid all these irritating human variables, multiple choice questions remain the dominant assessment model in ed tech. Far from being the latest, whizzy technology, multiple choice methodology was conceived around the start of the First World War and took off with automated test scoring machines developed by IBM in the 1930s.

Brillder creates ‘bricks’ – sophisticated investigations aimed at aspirational teens, authored by scholars, not by AI.

Whether they are described to you as DLUs (digital learning units) or OLUs (online learning units), the fact is that the vast majority of ed tech’s teach-and-test platforms use an educational approach which is about a century old. It’s digital now, true, and there might be wicked animations and sound effects, but be wary of multiple choice mutton being dressed as lamb.

It turns out – unsurprisingly – that the most productive ways to create educational content play to the strengths of human authors and the reach of technology. In primary education, computers can teach you how to spell or do basic numeracy. But as secondary education progresses, questions become more interesting and answers need to be constructed rather than regurgitated.

AI has fascinating capabilities in data crunching and operational efficiency. Pretty soon, everyone will use it in one way or another. But no AI I have encountered is capable of authoring educational content of any sophistication and, more surprisingly, AI is still hopeless at marking a written answer – a proper essay – to an open question like this: “Are electoral systems which are dominated by two parties sufficiently democratic?”

It’s the sort of question which you need to be studying politics or history at quite a challenging level – pre-tertiary (that is, sixth form) or tertiary – to even understand. To answer it well you do need to demonstrate considerable knowledge: the ability to compare a British general election to a US presidential election perhaps, and to have insight into how proportional representation works in countries like Germany or Italy. But, of course, there is no narrowly right or wrong answer here: arguments of equal value could be constructed using completely different data and perhaps coming to very different conclusions.

This is what most intelligent people call ‘real’ education. At Brillder, our achievement is to have devised a way of using technology which engages learners in sophisticated, interesting topics on all subjects – humanities as well as STEM. Our learning units (we call them ‘bricks’ because we’re all about building brilliant minds) help lay the foundations for high concept discussion, writing and problem solving. We don’t make the mistake of trying to displace teachers, but we have found a way of relieving workload because our superb resources are self-marking.

What we do at Brillder is use brilliant human authors – outstanding teachers and lecturers, recent doctorates, in some cases textbook authors – whom we ask to create the sorts of questions which test thinking as much as knowledge. In fact, the aim is to introduce knowledge through questions rather than merely to test what a student is supposed to have learned. We don’t waffle about AI or neuroscience, but we do insist on being interesting and challenging.

Technology makes our content dynamic, not intelligent. We use it to link the topic of our bricks to carefully chosen stimulus materials like videos or web pages; we use it for automated marking and for enabling authors to use 10 different question templates (of which multiple choice variations only represent two) alongside sound, images and equations. Tech also enables elements of gamification, like playing against the clock and replaying to improve your score. But we describe learning as a “serious game” – we don’t want to compromise on scholarship, we want to enliven it.

Our first customers are UK sixth form colleges and libraries but we have global ambitions and we have already been downloaded in 30 countries. We encourage any educators or investors intrigued by our philosophy to visit our website here: and explore our catalogue.

Our investor page is here: where you can also download our deck.

Don’t hesitate to contact us. We guarantee down-to-earth candour, free from the slightest hint of ed technical claptrap.

This article is sponsored content.