AI - The Greatest Equaliser Since the Printing Press?

I'm writing this newsletter for LinkedIn because I walked the dog.

Not a metaphor!

I walked the dog this morning, talked into Claude on my phone about what was interesting in the news this week, and why I feel optimistic about AI right now. Then I came home to a page of thoughts that wasn't blank, and reorganised and wrote them into what you're reading; I did this before my children woke up on Sunday morning. It's scheduled to go out on Monday morning while I'm in a client meeting.

This matters, and I'll come back to why.

But first: cost-benefit analysis. You don't need an economics degree to understand this concept; you use it every day without thinking about it. Every decision you make is a quiet calculation: is the value of doing this worth the cost of doing it? Time, money, energy, opportunity. When the cost is too high relative to the benefit, you don't do it.

Writing a newsletter the traditional way; sitting down with a blank page, drafting, redrafting, formatting, editing, publishing; is a cost-benefit calculation I would lose pretty much every week. I'm busy. I have fascinating and wonderful clients to serve, three children to raise, dogs to walk, a business to run, a platform to build, international standards to develop. If the only way to share these ideas was the old way, you wouldn't be reading this. The barrier to entry would be too high. The ideas would stay in my head, these conversations that happen on Linkedin reserved only for my closest friends and family.

But the barrier to entry has changed. And that change, the lowering of barriers to turning ideas into something the world can receive, is the most important thing happening in our economy right now thanks to AI. More important than any individual AI tool. More important than any model release or funding round. The barrier to entry for human ideas is collapsing. And the implications of that are enormous, and in my view; exciting.

The locked ideas

In 1714, the British government offered a prize of £20,000 (roughly £3 million today) to anyone who could solve the problem of determining longitude at sea. The solution came not from the Royal Observatory or the great astronomers of the age, but from John Harrison; a self-educated carpenter from Yorkshire with no formal scientific training. He spent 31 years building a series of increasingly precise clocks, fighting the establishment every step of the way, because the people who controlled the prize couldn't believe a working-class clockmaker had solved a problem that had defeated the finest minds in Europe. Harrison's idea was always there. It just had to fight through extraordinary barriers to reach the world.

Paul Romer won the Nobel Prize for showing that ideas are what he called "non-rival goods": unlike land or oil, an idea doesn't deplete when you share it. If I share an insight with you, I still have it. If a thousand people use it, it doesn't wear out. The more people who can create and share ideas, the more productive the whole economy becomes. It's the gift that keeps on giving.

In the 1950s, Robert Solow noticed something that puzzled economists for decades: when you added up all the traditional inputs to an economy; the land, the labour, the machinery, the capital; a huge chunk of growth was left unexplained. Where was it coming from? Solow called it "the residual," and it turned out to be the most powerful factor of all: knowledge. Ideas. The human capacity to solve problems in new ways. It's now understood that this knowledge capital; our collective ability to innovate; drives more economic growth than any other factor. And unlike a factory or a barrel of oil, it gets more powerful the more people who have access to it.

Which raises a question I keep turning over: if the most powerful driver of economic growth is the creation and distribution of ideas (it is), and if the tools for creating and distributing ideas have historically been available only to a narrow, educated, well-resourced subset of the population, how much innovation has the world been leaving on the table?

It seems to me sensible to think, a lot.

Think about what it actually takes to get an idea out of your head and into the world. You need to write it down coherently. Structure an argument. Draft a proposal. Format a pitch. Compose something people will actually read. These are skills that require years of education, years of practice, and the luxury of time. The ideas we see - the books, the businesses, the articles - are a vanishingly narrow subset of the ideas that actually exist. They're the ideas that happened to be held by people who also had the communication skills to express them.

Think about the stories that move us most. Malala was shot for wanting an education and became the youngest Nobel laureate. Oprah grew up in poverty in rural Mississippi and became one of the most influential communicators in history. @Tara Westover grew up in a survivalist family in Idaho with no formal education, taught herself enough to reach University of Cambridge and earn a PhD. If you haven't read Educated, read it; that story very nearly languished in rural Idaho and was never told.

These are beautiful stories. But they're beautiful precisely because they're rare. For every one, there are millions whose ideas never found an outlet. Not because they lacked but simply because the barriers were too high.

Barriers will always exist. But so many are being lowered right now, so fast, that I can't help thinking: imagine what's swilling around out there waiting to come through. Female-founded businesses. A teenager's app. Someone's blog that makes me reconsider something I was sure about. The ideas I now encounter on my LinkedIn feed from people I'd never have heard from two years ago; people who make me think, consider, question. Yes, there's slop. There's a lot of slop. But there are also glimmers of gold from minds that would previously have stayed invisible, and I find that enriching in a way I didn't expect.

A 12-year-old girl, and what access alone can't fix

When the internet arrived, the optimists said universal access to information would democratise education. Khan Academy , Wikipedia , Massachusetts Institute of Technology open courseware; all free, all there for anyone who wanted them. And not much changed in terms of inequality. Because the barrier wasn't just access. It was motivation. Information on the internet is like a library with the doors open; it only helps the people who walk in.

AI is different. Not because it provides information; the internet already does that. But because it adapts. It meets the learner where they are.

It can put maths into the language of football so a boy who couldn't engage with the subject now can. It can break learning into 20-minute blocks punctuated by physical tasks for the kid with ADHD who is effectively locked out of education because s/he can't sit still for an hour. It can encourage a child stuck at 9pm to try a different angle; be patient in a way that a tired parent at the end of a long day simply cannot.

The critical point: the alternative for these children is not a human tutor. The alternative is nothing. The realistic comparison isn't AI tutor vs human tutor. It's AI tutor vs no tutor at all. And thus, often, no or woefully inadequate education.

I see this in my own family. My youngest daughter is a hugely independent thinker who has no interest in doing what she's told by anyone, ever. School didn't work for her; the way it didn't work for Richard Branson , or Simone Biles , or any number of people who are on their own little planet and are wonderful for it. She's an entrepreneur through and through. But school said she was bottom set maths. If she were in school, she'd be in year 8 right now, barely clinging on to set 4. Instead, three weeks after she came out of school, she was teaching herself GCSE subjects with Claude and is now a good way through fourteen GCSE subjects on her own steam, including Maths, which at last test she's on track for 9's by next year. Three years ahead of schedule. Without AI; not possible. With AI, with the autonomy she and so many others require to engage; she's flying. But she must conform, I hear some say. But why? As soon as we leave school with a good education, we're expected to fill our own schedule in line with long term goals, set our own priorities, work towards long term tasks in a way that produces results, not sit in a room full of people doing exactly what we were commanded to focus on for that particular 40 minute period.

A World Bank randomised controlled trial in Edo State, Nigeria tested this use of AI in personalised education at scale. Students aged 12-17 used an AI tutor twice a week for six weeks. Learning gains equivalent to nearly two years of typical schooling. Outperformed 80% of all education interventions previously studied in the developing world. Girls, who initially lagged behind boys, made the most dramatic progress and closed the gap. Cost: roughly $5 per student. Many had never used a computer before.

Two years of learning in six weeks. That's not incremental, a little "should-we, shouldn't-we is it ethical or is it an unwanted intervention from the west"; this is structurally allowing sections of the population previously blocked from learning, to learn.

And it's not any old free ChatGPT subscription, by the way. It's a configured AI assistant, configured by someone who knows how to configure one.

AI lets us get back to the work that matters

Here's something that bothers me, a lot:

We weren't trained to be incredible thinkers so we could waste hours formatting slides on PowerPoint. How many hours have you spent over the course of your career messing around with formatting? We didn't earn degrees so we could move files around on SharePoint and spend 1.8 hours a day looking for information that should be at our fingertips. We didn't learn to think so we could try to get page numbers in the right corner, compress a PDF to be received by email, and format text nobody's going to read properly anyway.

(Incidentally, that 1.8 hours per day figure is the average time a knowledge worker spends looking for information in their organisation. A single well-designed AI assistant can remove almost all of it. It takes fewer than ten minutes for an AI-capable person to set up.)

If AI can allow teachers to spend more of those moments face to face with students; understanding their position, questioning them, supporting them, because the AI has dealt with the rote drudge; then let us all get back to teaching. Let us get away from the droning research and trawling through paperwork when we could be talking to our clients. Let us stop wasting time on writing documents from scratch that are the same structure each month, and instead take that extra thirty minutes to personalise the covering note, listen to what our clients actually need, have coffee with them, commune with them, break bread.

This week, after a client workshop, I could have spent half a day trying to remember what was said, deciphering my written notes, and building a document from scratch that sets out some fairly generic next steps. That's what I would have done two years ago. Instead, I spent that time interrogating the transcript from our conversation; stress-testing ideas about what's going to help them in hundreds of different ways at the touch of a button, researching what's worked elsewhere from billions of data sources no team of research assistants would access for me (not least at 9pm on a Friday, which is when I was doing this). I spent that time really understanding their problem and digging into the possible ways to address it; thinking about them, and formulating the solution that's actually going to work. Not formatting PowerPoints.

And that's what I want to be doing. That's where the value is. That's what we were trained for.

Cognitive decline? What cognitive decline?

I'd argue the opposite is happening. When you take the drudge away, what's left is the good stuff. The thinking. The connecting. The relationships. The creating. That moment of incredulity and delight that results from finally understanding what AI can do for us; how freeing, how liberating it is to leave behind the work that takes up 80% of our time and contributes zero to our satisfaction.

In psychology, there's a concept called the "Zeigarnik effect": unfinished tasks occupy mental bandwidth even when we're not working on them. That formatting we haven't done, that document we haven't finished, that information we can't find; it's all sitting in the back of our minds, consuming cognitive resources. Remove it, and the mind doesn't go blank. It goes to work on the things that actually matter. That's not cognitive decline. That's cognitive elevation.

Configuration: our greatest enemy or our greatest ally

None of this is to say this is the way AI is currently being used. No; we all see the slop.

AI can be used badly. We've all seen it. It's everywhere on LinkedIn; the flat, soulless posts that read like they were generated in 30 seconds by someone who couldn't be bothered to think. AI as a cheating machine for students who can't be bothered to do their homework. AI as a shortcut that bypasses thought rather than elevating it.

Here's an example I use a lot, because it illustrates the whole thing perfectly.

An HR team is running a hiring process. They ask AI to screen applicants at the first line. Just feed in the CVs, let it filter. What do we get? Bias. The AI reflects every pattern in the data it's been trained on, which means it's racist, sexist, and wrong in all the ways we've spent decades trying to correct. It becomes the enemy of our EDI policy. That's AI used badly: unthinkingly, unconfigured, as a blunt instrument.

Now take that same technology and configure it as a custom assistant instructed with reference to our EDI policy, our company values, the kind of team we want to build, the skills we're looking for, the growth we expect, our hiring strategy and priorities as a whole. That same piece of kit is now our EDI policy's greatest ally. It's our greatest ally in quickly finding the cream; the top 20% we now have time to have human eyes dwell on, pick over, speak with, question, challenge, connect.

Same technology. Opposite outcomes. The difference is configuration; how it was set up, what it was told, what it was for. That difference lies entirely in our hands.

This is AI capability. It's not what we think it is.

So the difference is configuration. Which raises the obvious question: who in our organisations knows how to do this?

If our people are left to their own devices with AI, we take pot luck on whether they're using it badly or well. Some will figure it out. But most won't (as we've seen haven't we, Linkedin folks!). And the ones who don't will either ignore it entirely or use it as a blunt instrument; the EDI-policy-enemy version, the flat-LinkedIn-post version, the "I asked ChatGPT and it said" version that makes everyone in the room wince.

But if we engineer for using it well; if we deliberately build the capability to configure, to contextualise, to direct AI toward genuine purpose; then everything we've been talking about in this newsletter becomes possible. The cognitive elevation. The barriers lowered. The cream rising from a much larger pool.

This is what AI capability actually means. It's not tool training; the tools change too fast for that to be a sensible investment. It's not "prompt engineering"; that's a narrow technical skill that's already being automated. It's something more fundamental.

It's the ability to ask: where does my human value lie, and how can I use the best tools available to extend it, scale it, nurture it, develop it?

For a teacher, that might mean: my value is in understanding each student and challenging them to think. How can AI handle the marking, the administration, the resource-finding, so I can spend more time doing that?

For a financial analyst: my value is in pattern recognition across complex systems. How can AI handle data gathering at scale and speed so I can spend more time on the analysis that creates insight?

For a leader: my value is in judgement, relationships, and strategic direction. How can AI handle more of the information retrieval, the first drafts, the reporting, so I can spend more time on the decisions and communication with my team that shapes the organisation and its trajectory?

That question; where does my human value lie, and how do I extend it; is the question. And the people we invite into our organisations to help with AI need to be able to show us how to answer it. Not how to use a specific tool. How to think about the whole thing.

Playing with ideas (and why this matters more than we might think)

I've always quite liked branding. I'm a metaphor person; I wrote my thesis on the metaphorical techniques used in economic debate, and how those metaphors shape policy. (Spoiler: a lot. George Lakoff's work on this is extraordinary if you haven't encountered it; metaphors aren't decorative, they're structural. The metaphor you choose determines the policy you build. "The economy is a machine" produces very different policy from "the economy is a garden." I could talk about this for hours, and sometimes do. Don't bait me.)

So I've always been interested in the idea that images shape meaning. There's a natural fit with branding, graphic design, visual communication. But I'm not graphic. I never learned Adobe. I'm not very artsy, is what I say.

What I do love is when a metaphor exists both in words and imagery. One of the people I admire most for this special kind of intelligence is Chris Walsh . What's been brilliant about AI is that I can play with these things now. I can play with the imagery I want to use to display my metaphorical structures. Before, I was reliant on my ability to articulate them, find and hire and pay the bill for a quality graphic designer, and iterate over a period of weeks. Now I can whip up an image and say to Chris: this is the kind of thing I'm thinking; that moment of glee. You see? And we've saved weeks of confusion. He can go spend his hours working his magic instead of producing endless decks of options for me to refuse, and I can go back to my business and await a beautiful result. Thanks, Midjourney .

That's another barrier lowered. Not a life-changing one like education in Nigeria. But a real one, in my working life, every day. And these things compound.

Build it, test it, learn, iterate

In technology, there's a concept called the waterfall method. It's what happens when a couple of smart people come up with an idea, hire a tech team for a million pounds to build it over a year according to specification, and discover at the end that nobody actually wanted it. Money gone. Time wasted.

The alternative is agile, the domain and expertise of a dear friend of mine Bazil Arden: build a minimum viable product (MVP) as quickly and cheaply as possible. Get it in front of real users. Gather feedback. Refine. Iterate. Rebuild. As one of my favourites, Bettina Briz , puts it: just get it out there and test. The faster and less expensively you can iterate, the better products you end up with. This much is now indisputable.

AI supercharges agile. A kid from Kennington with a great idea for an app can now build it; actually build it, with AI-assisted development tools, in hours rather than months, and for a grand total investment of a £20pm Claude ChatGPT Mastery or Claude licence. Test it with real users. Iterate based on feedback. If it's good, it gains traction. And then that kid walks into a conversation with an investor holding a working product and real user data (not an inadequate PowerPoint and an accent that minimises his chances of funding to effectively zero). That's a fundamentally different conversation: founders with traction and revenue have got something, no matter the accent.

I always liked that about being a swimmer rather than a gymnast or dancer: the clock doesn't lie. There's no bias involved in numbers. Traction and revenue are the same; it's hard to say there's no business when the pound signs say otherwise.

Women-founded businesses still receive less than 2% of global VC funding. Minority-founded businesses face similar odds. As a woman who has just completed a funding round, I can tell you the barriers are real. AI doesn't fix the bias. But it lowers the barrier to entry so dramatically that the gatekeepers become less relevant. If we can build our MVP instead of raising a seed round, we don't need permission to start.

Muhammad Yunus won the Nobel Prize for showing that people aren't poor because they lack ideas; they're poor because they lack access to capital. AI is doing something analogous for knowledge capital. Those tools are now available to everyone with an internet connection and £20-30 a month.

This is why digital access; wifi, devices, connectivity; should be at the very top of the policy agenda. This isn't a luxury or a nice to have; it's foundational equality. Locking people out of AI today is like locking them out of language itself.

The top 5% of 10,000

So, back to the dog walk.

You're reading this because AI lowered the barrier enough for me to pay it. Thank you, Claude. Thank you, LinkedIn scheduling. Thank you, Zapier , OpenAI , Google Gemini, for giving me space to build connections with people like my cofounder Louise Ballard (Moody) , with kindred spirits and professional allies like Robyn Agoston , Iliana Grosse-Buening , Dr Julia Stamm FRSA , Jeffrey Hughes , Dan Goldin , Sharon K. Gillenwater , Jeremy Utley , John O’Brien MBE Anthropist , Prof. Joanna Berry 🐝 , James Utting , Isabel Berwick , and so many others; connections that may never have been made without AI freeing me enough to include some idea-sharing in my start-up CEO daily life.

I want to be honest about something here as I expound the virtues of AI. I'm not optimistic because I see AI being used brilliantly everywhere. I don't. There's a lot of slop. A lot of lazy, unconfigured use that degrades rather than elevates. A lot of flat text from people who've outsourced their thinking rather than upgrading it. A lot of downright dangerous use that defies security and regulation and threatens businesses' integrity and reputation. We've all seen it, we continue to see it. The risks are real.

But I also see pockets of it being used well. Little glimmers of gold amidst the noise. And I can't help thinking: imagine if more people used it this way. Imagine a world where the ideas that reach us aren't just the cream of a tiny fraction that made it over sky-high barriers, but the cream of a vastly larger pool.

The top 5% of 10,000 is a very different proposition from the top 5% of 100.

The same principle applies to A/B testing, pitch deck screening, job applications, product development, research, education. When we lower the barrier, we don't just get more. We get better. Because the best of an A/B/C/D/E/F...Z run persistently over weeks is better than the best of just an A/B test. The best 10 pitch decks of 50,000 deserve the time and attention of our top people; and that 10 is better than the 10 from the 500 we used to review (inconsistently) without a first-line AI filter. The best products of 100,000 contenders are better than the best of only 1,000. The best ideas of 100,000 a week on LinkedIn are better than the best of only 1,000. And on.

I work with organisations across financial services, education, manufacturing, and government to build AI capability. I sit on AI Advisory Boards. I chair committees and contribute to them, one of which is committed to developing international standards for whether AI adoption is actually making people's lives better. And I'm also a mother of three who watches this technology change what's possible for her children every single day.

I'm not optimistic about AI because I see it being used brilliantly everywhere. I'm optimistic because I think it's necessary to be. It's necessary that we focus not just on the risks, but on the potential. That we watch not just the gaffes and laugh at the slop, but also notice, celebrate, and scale those glimmers of gold; those moments when someone's eyes light up because they've just understood what's now possible that wasn't before. That moment when they realise they can spend their time doing that top 10% of their job they really got into this gig for, and leave the other 90% to be picked up by well-configured AI.

And I think, if we're paying attention, and we look really hard at those glimmers, we're looking at the most significant expansion of human opportunity in several centuries.

Technology used well: ideas, thoughts, products, innovations, that would never have seen the light of day, reaching people who need them.

Technology used badly: algorithms that steal our children's attention, flatten discourse, connect the wrong people, exacerbate the ugly side of human nature.

That's a choice. And it's one for all of us; as leaders, as parents, as policymakers, as citizens.

To borrow from one of my boards:

If not us, then who.

If not now, then when.

This is newsletter #40. If you've been here since the beginning, thank you. If this is your first, welcome. And if you disagree with every word of this, I'd love to hear about that too. That's how ideas get better.