David Sheret

The first human-reaction to any truly-new technology isn’t strategy. It isn’t policy. It isn’t investment-analysis.

It’s a wee scrunch of the face.

A kid seeing a steam-locomotive for the first time doesn’t see a transport-platform. They see iron, smoke, heat and speed coming towards them. A family hearing a voice through a telephone for the first time; yip, we used to call them telephones, didn’t think about network-effects. They wondered, in some deep-old part of themselves, whether distance had been broken. When electricity entered the home, I suspect it wasn’t first understood as infrastructure. It was light-without-flame, power-without-muscle and danger-hidden-inside-a-wire.

Artificial-intelligence sits in that same emotional-history, I think. It’s new, but our reaction to it’s ancient. We’re excited because it seems to extend us. We’re scared because it seems to replace us. We look at the machine and ask two questions at once: what can I now do, and what might now be done to me?

That’s the honest starting-point. Not hype. Not doom. Just the old human-feeling of standing beside something powerful and not quite knowing whether it’s a servant, a rival, a tool, a mirror or a warning.

AI isn’t simply a software-story anymore. It’s an economic-story, a labour-story, a media-story, a trust-story, a power-story and, in a wee bit of a strange way, a spiritual-story. It asks what makes a person useful. It asks what counts as creativity. It asks who owns knowledge once knowledge can be scraped, trained, summarised and sold back to us. It asks whether judgement can be automated, or whether judgement becomes more precious when everything else becomes easier.

We’ve been here before, although never quite like this.

The printing-press didn’t merely make books cheaper. It disturbed authority. It moved ideas faster than institutions could contain them. The steam-engine didn’t merely move goods. It remade work, cities, time and empire. The telegraph collapsed distance. Radio put voices inside homes and made politics intimate. Television turned image into power. The internet promised a democratic-library and delivered email, e-commerce, scams, revolutions, surveillance, podcasts, cat-videos and the largest attention-machine ever built.

Every major-technology seems to follow the same emotional-arc. First wonder. Then over-claim. Then panic. Then money. Then disappointment. Then regulation. Then ordinary-life. What once looked like magic becomes, well, kinda plumbing.

Currently, I see AI sitting somewhere between over-claim and plumbing.

That makes it difficult to talk about properly. The language around it’s already overheated. Every product’s “transformational”. Every assistant’s a “co-worker”. Every chatbot’s an “agent”. Every marginal-efficiency gain’s presented as a revolution. It’s tempting to roll your eyes. Sometimes, frankly, I think we should. It’s kinda healthy and we Scots are great at it. 

But hype doesn’t prove emptiness. The railway-boom had frauds and fortunes, but railways still changed the world. The dot-com bubble destroyed capital, but it also helped build the platform-economy. Crypto was wildly over-sold, but it left behind serious arguments about verification, digital-ownership and trust; indeed they’re ongoing and some would say even less clear. AI is kinda the same. There’ll be waste, vanity, overvaluation and nonsense. There’ll also be real change.

Like any commodity in the current-world, especially in the Western-world and particularly in markets shaped by the United States, AI attracts hyperbole like nobody’s business. That isn’t an accident. It’s part of the system.

It’s too simple, and kinda lazy, to say that private-equity, venture-capital and big-technology companies are just selling stories. But story-selling’s absolutely part of it. Capital doesn’t move on spreadsheets alone. It moves on belief. Belief becomes investment. Investment becomes exposure. Exposure becomes risk. Risk needs to be managed. And one of the oldest ways of managing risk’s to persuade more people to believe in the thing you’ve already backed.

That’s why the AI-story matters commercially. The story supports the valuation. The valuation supports the next raise. The next raise supports the hiring-plan. The hiring-plan supports the product-roadmap. The product-roadmap supports the next story.

None of that means the technology’s fake. It means the market’s human.

And that, perhaps, is the point we keep missing. AI is often discussed as if it’s arrived from outside society, when in truth it’s arrived through people. It carries our incentives, our ambitions, our blind-spots, our shortcuts, our greed, our brilliance and our fear.

Anyone who’s used AI seriously knows the duality. It can draft a contract-note, test a marketing-line, summarise a tender, write code, analyse a spreadsheet, translate a document, generate an image, suggest a melody and give a small-business capabilities that once belonged only to large-firms. It can also hallucinate, flatter, misread, invent sources, reproduce bias and produce absolute-nonsense with the deluded confidence of Chewin’ The Fat’s Big Jock in the 19th. 

That duality isn’t a bug in the story. It’s the story.

Where AI goes next is becoming clearer. It’s moving from answer-machine to action-machine. The first public-wave was conversational: ask a question, receive a response. The next-wave is agentic: give a goal, and the system plans steps, calls tools, moves data, books meetings, updates systems, drafts replies, compares documents and reports back.

That sounds useful. It is useful. But it also changes the risk. A chatbot that says something wrong’s one problem. An AI-agent that does something wrong’s another.

So the question isn’t whether AI can write a passable email. It can. The question’s whether it can safely enter the bloodstream of a company. Can it chase invoices without offending customers? Come to think of it, can humans do that? Can it review contracts without missing the clause that matters? Can humans do that? Can it triage patients without embedding old-bias in new-software? Can it assist lawyers without creating false-confidence? Can it support teachers without thinning-out human-attention? Can it help artists without stripping artists of value?

For me, there aren’t clean answers. There’re only better questions.

In law, AI won’t replace judgement, but it’ll devour routine-research. In medicine, it won’t replace compassion, but it’ll help read scans, surface patterns and challenge tired assumptions. In energy, it’ll optimise assets, model maintenance and manage complex-systems. In education, it may become tutor, examiner, cheat-code and equaliser all at once. In music, film and design, it’ll flood the world with competent-mediocrity and make those that use it in a creative way with true-feeling and purpose more valuable, not less.

That may be the most interesting cultural-shift. When production becomes easier, judgement becomes scarcer. When everyone can publish, editing matters. When everyone can generate an image, taste matters. When everyone can make a song, feeling and purpose matters. When everyone can ask AI for strategy, execution matters.

The premium moves from access to discernment.

Online-platforms are changing at the same time. Search is becoming answer-first rather than link-first. That sounds like a design-choice, but it may be a constitutional-change for the web. For years, websites, publishers, bloggers, companies and experts created material partly because search sent people to them. If the platform now reads the web, summarises the answer and keeps the user, the old-bargain weakens.

A local-newspaper report, a restaurant-review, a niche-technical blog or a thoughtful explainer may still feed the machine, but not receive the visit. The internet becomes less like a public-square and more like a set of giant reception-desks summarising what the public-square used to say.

Convenient? Absolutely.

A wee bit extractive? I think yes there too.

Meanwhile, trust’s already fragile. People don’t simply “consume media” anymore. They follow a person, a clip, a feed, a channel, a podcast, a group-chat, a creator, a mood. News arrives wrapped in personality. Politics arrives as performance. Expertise competes with confidence. Into that world comes synthetic-media: fake voices, fake images, fake video, fake intimacy, fake outrage, fake proof.

The danger isn’t only that lies become easier to make. It’s that truth becomes more tiring to prove.

And still, we’ll be excited.

We’ll be excited because a teenager in Inverness can make a film that once required a studio. A dyslexic-founder can write with confidence. A teacher can make personalised-learning materials in minutes. A small-consultancy can analyse markets with tools once reserved for multinationals. A nurse can get decision-support. A musician can sketch an arrangement. A scientist can explore chemical-space. A lonely or nervous person can rehearse a difficult-conversation before having it in real-life.

That isn’t trivial. Human-flourishing often begins with capability. A tool that gives more people more capability deserves optimism.

But optimism-without-seriousness is just marketing.

AI’s physical-footprint is really important, least it is to me. The cloud isn’t actually a cloud. It’s land, water, chips, cooling, substations, power-purchase agreements and grid-pressure. In-a-way, AI’s the most modern thing imaginable and one of the oldest things imaginable: a new hunger for power.

Governance matters too. Not because regulation’s always wise. It often isn’t. Some of it’ll be clumsy. Some of it’ll be late. Some of it’ll be written by people who don’t fully understand the thing they’re trying to control. But when technology starts influencing work, credit, education, policing, health, media and democratic-discourse, rules become inevitable.

Therefore, from where I sit, I think the public-mood is not really confused. It’s balanced. People are excited and worried because both reactions are rational. They remember that platforms promised connection and delivered addiction. They remember that globalisation promised efficiency and delivered insecurity for many communities. They remember that “move-fast-and-break-things” often meant breaking somebody else’s things.

So the task isn’t to choose between fear and excitement. It’s to earn the excitement and respect the fear.

We should welcome AI that helps doctors, teachers, carers, engineers, artists, small-firms and public-services. We should question AI that concentrates power, strips creators of value, automates discrimination, floods public-life with synthetic-sludge or turns work into a surveillance-dashboard. We should celebrate tools that help us think. We should be wary of tools that invite us to stop thinking.

The mature-position is neither boosterism nor doom. I think it’s stewardship.

AI will become more capable, more embedded, more personal, more regulated and more invisible. It’ll move from browsers into operating-systems, cars, hospitals, factories, classrooms, cameras, glasses, call-centres, studios and homes. It’ll be in the workflow before we notice it as a product. Like electricity, it’ll disappear into usefulness. Like electricity, it’ll still be dangerous when badly-wired.

That’s why I think that AI will continue to thrill and terrify us. Not because the machine’s alien, but because it’s familiar. It reflects our ambition and our laziness, our generosity and our greed, our creativity and our appetite for shortcuts.

AI doesn’t arrive from nowhere. It arrives into us.

And perhaps the real question isn’t whether artificial-intelligence becomes more human.

It’s whether we become more thoughtful.

What do you think?