The Dam, the Washing Machine, and the Ferrari

I argued this week at the The London School of Economics and Political Science (LSE) for She Shapes AI that AI will lead to an era of unprecedented abundance and human flourishing. Here is my 3 minute case.

Will AI automatically lead to human flourishing?

No. Not automatically.

But will it? Yes.

I've been helping organisations navigate tech transitions since the 2000s. Cloud, digital, social media; every time the rules changed, my phone rang. I am a human capital economist, specialised in Applied AI, and I advise governments, regulators, education, and corporates large and small on how to adopt AI. I also have three children, the neurodivergent one of whom is happily teaching herself fourteen GCSEs two years early through AI and a stack of books.

When I say AI leads to human flourishing, I'm living it. Here's why, in three pictures.

The dam.

Paul Romer won the Nobel Prize for proving that ideas are the engine of all economic growth. But for most of history, ideas have been trapped: behind limitations of education, capital, connections, race, geography.

AI has opened the floodgates.

A twelve-year-old girl in Nigeria with an AI tutor gained two years of learning in six weeks. For five dollars. Her alternative was not a human tutor; her alternative was nothing.

A founder without capital, connections, or a golf club membership can test a business or build an app for twenty pounds a month without permission from a VC.

The top 5% of 10,000 ideas is simply better than the top 5% of 100.

That is flourishing already.

PwC estimates AI could add $15 trillion to the global economy, and economic flourishing - the lift out of poverty towards abundance - is a chunky part of flourishing. We are capturing less than 5%, not because the technology doesn't work, but because we haven't trained the humans to apply it properly.

The washing machine.

I put mine on every day. But make no mistake, we have lost things because of it. Women used to congregate at the river with washboards; they built community, strong forearms, camaraderie. I can't wash clothes by hand. I also can't navigate by the stars, and neither can my kids. We have lost skills that were once important to people, protected, valued.

But every labour-saving technology trades one way of living for a different way of living.

When a teacher freed from paperwork spends time face to face with a child; when a nurse holds the hand of a patient instead of ticking boxes; when I get to spend my evening thinking about my client's needs instead of formatting PowerPoints; labour saving technology can - when correctly applied - be a force for real, human, good.

What we gain must simply be greater than what we lose.

In medicine, that case is already made. The first entirely AI-designed drug has passed human clinical trials. AI-assisted screening catches 29% more breast cancers. AI-designed antibiotics are fighting infections that kill five million people a year. Not projections; published, peer-reviewed results.

On the environment, AI-driven energy management saves an estimated 300 terawatt hours of electricity annually; roughly what Australia and New Zealand generate combined. Up to 175 gigawatts of additional grid capacity can be unlocked from existing power lines, without building a single new pylon. Google reports a 26:1 ratio: for every tonne of carbon its AI produces, its AI products help reduce 26 tonnes elsewhere. The energy cost of data centres is real. The net direction of travel is positive.

The Ferrari.

Will AI necessarily make us flourish? That is like asking whether dropping an F1 car on your doorstep makes you an F1 champion. Unprepared, no training, no team? We'll likely wrap it around the first lamppost on the corner. But give us all training, a team, practice, and clear directions to the track, and we might end up with a great many champions.

We've all seen both: the AI car crashes, the slop, the mechanical glowing blue images on LinkedIn. But we've also seen the gold. The difference is not the car; it's who is driving. And teaching people to drive is a big job, and it's not easy, but it is achievable.

We are standing in front of the greatest accumulation of human knowledge and capability ever assembled in one place. The barriers are down, the tools are here.

How can we NOT make a success of this?

The Crucial Distinction: What didn't fit in 3 minutes

Badly used AI and well used AI are two different tools.

Use unconfigured AI to screen job applications and it will reproduce every bias in its training data. Racist, sexist, ageist. The headlines are accurate. Now equip the same tool with your EDI policy, your values, the skills profile of the team, and strict instructions to screen in line with that context.

It becomes the most powerful ally your diversity policy has ever had; the first tool capable of enacting it consistently, at scale, without the unconscious biases every human interviewer carries into every room. Same technology, configured differently.

One is a liability. The other is a force for equity.

My twelve-year-old is autistic and ADHD. She was effectively excluded from school for not fitting the system. She is now teaching herself fourteen GCSEs two years early, and she is the most engrossed learner I have ever met. Her AI does not do one scrap of work for her. It guides her reading through our family library from Animal Farm to Communist Russia, to the Cold War, to geopolitical tensions, to Marx, to mythical origins of J.K. Rowling's beasts in Harry Potter; it directs her toward learning objectives, asks for reflections, widens her understanding, tests her knowledge, tracks her progress against over 400 objectives, and makes it visible to me daily in real time. She reads, thinks, writes, and creates extraordinary illustrated manuscripts by hand. The AI is the scaffold; she is the architect. Her sister listens to cell biology as teen-style serialised podcasts while cycling to sports training, learning GCSE science through a series we built together in an evening called Aria's Diary, designed to make complex science feel like a story you want the next episode of. Her little brother has his designing his reading programme to correspond to his running training and his need to move as a 10 year old boy.

Meanwhile, in schools across this country, children are using AI at every level to do the work for them, while educators and government examine the ethics and pretend it is not happening.

These are two entirely different things.

The people shaping AI policy do not, for the most part, understand the distinction.

The ethics of waiting.

We spend a great deal of time discussing the ethics of deploying AI. I would like us to consider the ethics of not deploying it.

When AI can deliver education to a child who has no access to one, where are the ethics in withholding it?

When it can save 300 terawatt hours of energy a year, where are the ethics in not deploying it?

When it can coordinate patient records and deliver better NHS care at a fraction of the cost to the taxpayer, who is asking about the ethics of delay?

Behavioural economists call this omission bias: the tendency to judge harmful actions as worse than equally harmful inactions. The risks of doing something feel vivid. The risks of doing nothing are invisible, because the people harmed by inaction are invisible. The girl who never got educated. The cancer caught too late. The energy never saved. These harms are real. But because nobody made a decision that caused them, nobody feels responsible.

Those in charge need to get informed

We need people shaping AI guidance who understand the difference between AI used badly and AI used well, because these are two entirely different things and the distinction is everything. And we need to help people configure AI for their purposes, not the technology companies' purposes. That distance is not technically complicated; it is the single highest-value capability investment any organisation or government can make right now.

Pointing out risks from the sidelines is easy. Quite frankly, it is the lazy route; it is always easier to articulate why something might go wrong than to do the hard, demanding work of figuring out how to make it go right. Choosing inaction when the tools to help are already here is not caution. It is a decision, and the people bearing the consequences are not the ones making it.

Can we do better?

I made this case at the LSE yesterday. The room was ultimately persuaded by the other side: that capitalism will do what capitalism always does, that tech companies will maximise for profit, that lawmakers will do nothing, and people will watch AI roll out like a bad train crash that makes the social media debacle look like pre-amble.

I understand the cynicism; I share it.

But this is exactly the self-fulfilling prophecy that concerns me most.

If we decide it can't work, we stop trying, and we guarantee the outcome we feared. That is what happened with social media. Our children are still paying the price. Our environment, our NHS, our society.

If we are standing in front of the greatest accumulation of human knowledge ever assembled and we cannot make a success of it, what does that say about us?

I refuse to believe that is the answer. I would rather be wrong and have tried than be right and have done nothing.

Don't kick sand from the sidelines. Join us.

We run a non-profit initiative called Wise Five, backed by practitioners, educators, technologists, and organisations across the country, built to guide this transition toward an outcome that works for everyone. If you can offer time, expertise, or network, get in touch. This is not about ego, profit, or power; it's about what kind of country we want to be - and we'd love to have you.

mackenzie@atheni.ai

Mackenzie Howe is CEO of Atheni.ai. She helps organisations adopt AI in a way that works for business and people.