Scotland Must Not Build a Digital Future It Cannot Escape

AI data centres are being promoted as engines of growth, innovation and green technology. Scotland should ask a harder question before giving them land, water, power and planning support: what happens when daily life, public services and human work are moved inside systems ordinary people cannot control?

Scotland is being asked to welcome a new kind of infrastructure. It will not look like a hospital, a school, a harbour, a railway or a farm. It will look like a large industrial campus, filled with servers, cooling systems, substations, security fencing, cables and backup power.

The language around it will be polished. It will speak of artificial intelligence, cloud computing, renewable power, economic opportunity, digital transformation, green growth, jobs and global competitiveness.

But a hyperscale AI data centre is not just a building. It is the physical machinery that allows public services, private commerce, banking, health care, education, transport, communication, policing, military analysis, advertising, employment systems and ordinary personal life to be moved into digital platforms.

That question now facing Scotland is whether Scotland is building a digital future that serves people, or one that people will be forced to enter because the non digital world has been quietly dismantled.

The UK Government has described data centres as “the engines of the AI age” in its AI Opportunities Action Plan. It has also designated data centres as Critical National Infrastructure, placing them alongside systems such as energy and water because they underpin economic activity and public services. That recognition tells us two things at once. Data centres are now central to modern life. They are also a point of national vulnerability.

For Scotland, the issue is no longer abstract. Proposed hyperscale data centres, including the 600MW Cato project near Auchtertool in Fife, show the scale of electricity, land and planning pressure that AI infrastructure can bring. The wider question is what these machines will process, who will control the systems, which jobs will be reduced, what services will depend on them, and what remains for people who cannot or will not live entirely online.

DANGER 1: DIGITAL LIFE CAN BECOME COMPULSORY

The first danger is not that a person chooses to use a digital service. The danger is that every alternative disappears.

A bank closes its branch and tells customers to use the app. A council closes a counter and sends residents to an online portal. A hospital sends appointment letters through digital systems. A railway requires a QR code. A shop refuses cash. A job application requires automated screening. A public service requires identity verification through a phone. A benefits system assumes everyone has broadband, a device, memory, literacy, confidence and a stable life.

In that world, digital access is not convenience. It is citizenship.

Audit Scotland has already warned that as more public and private services move online, people who do not use or struggle with digital technology risk being unable to access the services they need. The Scottish Government’s own digital vision acknowledges that tackling digital exclusion works best when it is place based and involves local communities.

That should be the starting point for Scotland. Digital systems must not become the only door into public life. Every essential public service should keep a real human route: a telephone number answered by a person, a staffed local access point, paper forms where needed, postal options, and support for people without smartphones or reliable internet.

The principle should be plain. No Scot should lose access to health care, housing, banking, welfare, transport, education, justice or local services because they cannot pass through a screen.

DANGER 2: DATA CENTRES PROCESS WORK, NOT JUST DATA

The second danger is that AI data centres are presented as neutral infrastructure when they are also machines for absorbing human work.

They will process tax queries, benefit claims, medical notes, planning applications, insurance claims, customer complaints, legal documents, school administration, job applications, warehouse routes, banking checks, translation, copywriting, software code, surveillance footage, military data, transport schedules and personal records.

That work is not imaginary. Much of it is currently done by people.

The jobs most exposed are not only factory jobs. They include clerical workers, call centre staff, junior legal staff, translators, customer service workers, insurance handlers, bank staff, writers, designers, junior coders, schedulers, public administration workers, retail support staff, HR screeners, compliance assistants and people who sit in the middle of ordinary office life.

Yet the Tony Blair Institute has estimated that full AI adoption by UK firms could save almost a quarter of private sector workforce time, equivalent to the annual output of six million workers.

Scotland cannot assume those forces will stop at Gretna.

The Scottish labour market includes public administration, banking, insurance, health administration, universities, creative work, customer service, transport, logistics, retail, legal services, energy, planning and local government. Those are precisely the sectors where AI can absorb routine tasks.

That does not mean every job disappears. It means many jobs are thinned. One person does the work of three. A team becomes a subscription. The first draft, first check, first call, first summary, first decision and first rejection are automated. Then a manager calls it productivity, because one must keep spirits up in the quarterly report.

Scotland should require every major AI data centre proposal to include an employment impact statement.

Not vague promises about construction jobs. Not glittering phrases about innovation. A proper statement should say which sectors the facility is expected to serve, how many permanent jobs will be created locally, what kind of work its AI customers are likely to automate, and what training or transition support will be funded for affected workers.

DANGER 3: GOVERNMENT CAN BECOME A PORTAL WITH NO COUNTER

The third danger is public power moving behind systems that ordinary people cannot question.

Government AI and cloud systems may process tax, welfare, housing, planning, immigration, policing, courts, social care, child protection, environmental regulation and emergency planning. Some of this may improve speed and consistency. Some of it may reduce delay. But a public service must remain accountable to the public.

If an algorithm scores a benefit claim, screens a housing application, sorts a planning objection, flags a family for review, ranks a job applicant or summarises a court document, citizens need to know how the system is being used, who checked it, what data fed it, how errors are corrected, and how a person can challenge the outcome.

A digital state can become very efficient at saying no.

Scotland should require a public register of automated decision systems used by devolved public bodies, councils and public contractors. It should name the supplier, the purpose, the data used, whether AI is involved, whether the output affects rights or services, and what human appeal route exists.

No public body should be allowed to hide behind the phrase “the system says”. The system is not sovereign. The citizen is not a nuisance in the machine.

DANGER 4: PRIVATE COMPANIES MAY CONTROL THE PUBLIC NERVOUS SYSTEM

The fourth danger is control.

The digital world is not stored in the clouds. It is stored in buildings, servers, cables, contracts and companies.

The key commercial actors in the global AI and cloud economy include Amazon Web Services, Microsoft Azure, Google Cloud, Oracle, IBM, Nvidia, OpenAI, Meta, Apple, Salesforce, Palantir and many sector specific firms in health, finance, education, defence, retail and logistics. Governments may set policy, but much of the infrastructure is privately owned or privately operated.

That creates a constitutional question for Scotland. What happens when public services depend on private platforms? What happens when a supplier changes prices, changes terms, withdraws a service, suffers a breach, moves data, restricts access or becomes entangled with another government’s legal demands?

This is not a call to reject all private technology. It is a call for public control over essential dependency.

Scotland should insist that public services using AI and cloud systems have contract transparency, data residency clarity, exit plans, backup providers, human fallback routes and continuity plans. If a service cannot operate when one cloud supplier fails, it is not resilient. It is merely confident on a sunny day.

DANGER 5: IF THE POWER FAILS, DIGITAL SOCIETY FAILS WITH IT

The fifth danger is the one everyone understands once they say it out loud.

A digital only society depends on electricity.

If power fails, data centres fail unless backup systems hold. If networks fail, services fail. If cloud access fails, offices may not function. If card payment systems fail, shops may not sell food. If fuel payment systems fail, vehicles may not move. If hospital records are inaccessible, care is delayed. If transport systems cannot communicate, travel slows or stops. If councils cannot open case files, vulnerable people may wait. If phones and internet go down, families cannot reach help.

The UK Government’s own data centre policy says data centres host and support modern life, from patient records and emails to product data and financial systems, and are critical to nearly all economic activity and public services. That is a remarkable admission. It means the modern economy has a shared point of fragility.

Scotland should treat digital resilience like water, food and heat.

Every council, NHS board, transport body, emergency service and essential public agency should have analogue continuity plans. Printed emergency contacts. Local radio plans. Paper forms. Offline payment contingencies. Community resilience hubs. Backup communications. Staff trained to operate without cloud systems. Critical records available in secure offline form where lawful and safe.

This is survival planning.

A society that has removed cash, branches, counters, printed records, paper forms, staffed phones and local offices cannot simply return to normal when systems fail. It has destroyed the bridge back.

DANGER 6: CASHLESS SCOTLAND WOULD BE A FRAGILE SCOTLAND

Banking and payment systems deserve special attention.

The Financial Conduct Authority introduced access to cash rules in 2024, requiring banks and building societies to protect access to cash. That helps with access to withdrawals and deposits. It does not by itself guarantee that every essential business must accept cash.

A person may have cash in their hand and still be unable to use it if shops, transport or services refuse it. During outages, cyber incidents or banking disruption, cash can become the simplest backup system in the country.

Scotland should push for an essential cash acceptance standard.

Food, fuel, medicines, basic transport and critical local services should have a way to accept cash during digital disruption. Businesses should not be punished for modernising, but essential life should not depend entirely on card terminals and apps.

Cash is not merely money. It is offline public infrastructure.

DANGER 7: MILITARY AND SECURITY USES WILL NOT ALWAYS BE VISIBLE

AI data centres may also process defence, intelligence and policing workloads. These can include satellite imagery, drone feeds, cyber defence, maritime tracking, border risk systems, facial recognition, biometric matching, signals intelligence, logistics, disinformation tracking, missile defence modelling and autonomous systems.

Some of that work may be necessary for national security. But secrecy and automation are a dangerous combination.

Scotland already hosts infrastructure connected to defence, North Atlantic security, aviation, space and maritime surveillance. As AI and cloud systems become more central to military and security operations, Scotland should know whether data centres, networks, launch sites or energy infrastructure are serving civilian, commercial, defence or dual use purposes.

That does not mean every sensitive detail can be public. It means democratic oversight must not be replaced by commercial confidentiality and national security fog. There is fog enough in Scotland without importing more at public expense.

THE HUMAN OPT OUT MUST BE DESIGNED NOW

The human opt out cannot mean refusing all technology. Most people will use digital services where they are useful. The opt out means something more basic.

It means the right to live, bank, travel, access health care, use public services, receive letters, speak to a human, pay for essentials, prove identity, apply for help, complain and participate in civic life without being forced into a single digital route.

Scotland should create a Digital Choice Standard for essential services.

That standard should require:

A human telephone route for essential public services.
A paper or assisted application route for benefits, housing, health, justice, education and local services.
A staffed access point within reasonable distance or through local community hubs.
A cash or offline payment contingency for essentials.
A right to request human review of automated decisions.
A clear statement when AI is used in a public process.
A ban on digital only access for essential devolved services unless an assisted alternative is available.
Regular testing of public services under power, internet and cloud outage conditions.

This is how Scotland can navigate the digital age without surrendering to it. The aim is not to stop technology. The aim is to keep humans above the system.

WHAT SCOTLAND SHOULD DEMAND BEFORE APPROVING HYPERSCALE AI DATA CENTRES

Before Scotland gives land, water, power, grid capacity or planning support to hyperscale AI infrastructure, it should demand plain answers.

Who owns the data centre?
Who are the likely customers?
Will the workloads be government, commercial, military, health, finance, education, advertising, surveillance, cloud hosting or AI training?
Which public services may depend on it?
Which companies will profit?
How much electricity will it draw at full operation?
What happens to local grid capacity?
Will public money, grants, tax breaks, discounted electricity or planning priority be involved?
How much water will it use?
What backup generation will it require?
How many permanent local jobs will be created?
Which kinds of Scottish work may be automated by the systems it hosts?
What non digital routes will remain for affected public services?
What happens when the power fails?
What happens when the provider changes the terms?
What happens when a citizen says no?

These questions are not anti technology. They are pro public life. A society that cannot answer those questions is not planning a digital future. It is signing one. Scotland does not need to fear technology. It needs to govern it.

What exactly will the AI data centres process?

GOVERNMENT AND PUBLIC ADMINISTRATION

This includes tax records, benefit claims, housing applications, planning systems, licensing, passports, immigration files, public consultations, grant applications, council contact centres, procurement, court documents, prison management, social care files, child protection records, environmental permits and emergency planning.

These systems are controlled by central government departments, devolved administrations, local authorities, courts, agencies and contractors. In the UK, that means bodies such as HMRC, the Department for Work and Pensions, the Home Office, the Ministry of Justice, local councils, NHS bodies, the Scottish Government and public contractors.

The work at risk includes clerical administration, call handling, document checking, case triage, routine correspondence, form processing and first line public contact.

For citizens, the danger is that access to public life becomes conditional on digital compliance. If a person cannot use the portal, remember the password, pass the identity check or reach a human being, they can become administratively invisible. The form does not need to dislike you. It merely needs to refuse you.

HEALTH AND NHS SYSTEMS

This includes medical scans, clinical notes, appointment systems, hospital records, discharge letters, ambulance demand, NHS call triage, prescriptions, medical coding, remote monitoring, genomics, drug discovery, clinical trials and patient risk prediction.

These workloads are controlled by NHS boards, health departments, universities, pharmaceutical companies, medical technology firms, cloud providers and private contractors. Major commercial actors in this space include Microsoft, Amazon Web Services, Google Cloud, Oracle, Palantir, IBM, Epic, Cerner, Philips, Siemens Healthineers and large pharmaceutical companies.

The work at risk includes transcription, coding, appointment scheduling, administrative triage, parts of diagnostic support, call handling and routine record management.

The public risk is not only job loss. It is loss of access. A health service that cannot function without cloud records, AI triage and digital booking becomes fragile when systems go down. A paper file in a cabinet was slow. A locked cloud system may be absolute.

MILITARY, INTELLIGENCE AND SECURITY

This includes satellite images, drone feeds, maritime tracking, cyber defence, signals intelligence, border systems, facial recognition, biometric matching, battlefield logistics, weapons testing data, missile defence, autonomous systems, target recognition, disinformation tracking and critical infrastructure monitoring.

These workloads are controlled by military bodies, intelligence agencies, defence ministries, police bodies, NATO structures, defence contractors and major cloud platforms. Relevant actors include the UK Ministry of Defence, the US Department of Defense, NATO, GCHQ, the National Cyber Security Centre, police forces, BAE Systems, Lockheed Martin, Northrop Grumman, Raytheon, Palantir, Anduril, Microsoft, Amazon Web Services and Google Cloud.

The work at risk includes analyst support, mapping, monitoring, logistics planning, cyber triage, image review and intelligence sorting.

The public risk is secrecy combined with automation. Once defence and security decisions are processed through machine systems, the public may never know what data was used, who reviewed it, or whether an error was corrected before action followed.

BANKING, PAYMENTS AND INSURANCE

This includes card payments, online banking, mortgage screening, credit scoring, fraud detection, insurance claims, pensions, investment research, anti money laundering checks, payroll systems, tax preparation, audit sampling, compliance monitoring and customer service.

These workloads are controlled by banks, insurers, pension funds, accountancy firms, payment processors, regulators and cloud providers. Key companies include Visa, Mastercard, PayPal, Stripe, Worldpay, Barclays, Lloyds, NatWest, HSBC, Aviva, Legal & General, BlackRock, Deloitte, PwC, EY, KPMG, Microsoft, Amazon Web Services and Google Cloud.

The work at risk includes bank branch roles, call centres, claims handling, clerical finance, junior analysis, bookkeeping, audit support and routine compliance work.

The dependency risk is obvious. If digital payment systems fail, many people cannot buy food, fuel or medicine. Cash has been treated as old fashioned. In a power cut, it begins to look like civilisation.

COMMERCE, RETAIL AND LOGISTICS

This includes online shopping, stock forecasting, delivery routing, warehouse automation, customer complaints, product recommendations, personalised pricing, payment disputes, returns, supplier risk, advertising, inventory systems and workforce scheduling.

These workloads are controlled by private corporations, supermarkets, logistics firms, online platforms, payment processors and cloud providers. Key companies include Amazon, Tesco, Sainsbury’s, Asda, Morrisons, Ocado, DHL, DPD, UPS, FedEx, Royal Mail, Shopify, Salesforce, Microsoft, Oracle and Google Cloud.

The work at risk includes retail administration, warehouse planning, customer service, call centres, stock control, dispatch, marketing support and junior buying roles.

The public risk is supply chain fragility. Remove electricity and it fails.

EDUCATION AND UNIVERSITIES

This includes lesson planning, marking support, admissions, student records, plagiarism detection, online tutoring, university research, grant applications, lecture transcription, learning analytics, exam administration and student support chatbots.

These workloads are controlled by schools, councils, universities, exam bodies, education technology firms and cloud providers. Key actors include local authorities, the Scottish Qualifications Authority or its successor bodies, universities, Microsoft, Google, Turnitin, Pearson, Canvas, Blackboard and education technology platforms.

The work at risk includes administrative support, tutoring, marking assistance, admissions processing, research assistance, translation and some junior academic support.

The social risk is that education becomes less human around the edges. The teacher may remain, but the support system around pupils can be hollowed out and replaced with dashboards. A child cannot be raised by a dashboard.

MEDIA, CULTURE AND CREATIVE WORK

This includes news aggregation, article summaries, social media posts, advertising copy, image generation, video editing, music production, voiceovers, translation, dubbing, book summaries, design concepts, game assets and audience analytics.

These workloads are controlled by publishers, broadcasters, platforms, agencies, studios, music companies, streaming services and AI firms. Key companies include Meta, Google, YouTube, TikTok, Spotify, Netflix, Disney, Adobe, OpenAI, Microsoft, Amazon, Apple, Canva and major advertising networks.

The work at risk includes junior writing, copywriting, translation, design, editing, voice work, production assistance, marketing content, stock photography and routine publishing tasks.

The cultural risk is not merely unemployment. It is the flooding of public life with synthetic material until real work, original reporting, local art and human judgement are buried under cheap output. A country can lose its culture by drowning it in sludge.

LEGAL AND PROFESSIONAL SERVICES

This includes contract review, legal research, disclosure, due diligence, compliance checks, board papers, procurement law, intellectual property searches, corporate filings, tax advice, audit work and regulatory reporting.

These workloads are controlled by law firms, corporate legal departments, accountancy firms, regulators, banks, insurers and cloud providers. Key companies include Thomson Reuters, LexisNexis, RELX, Microsoft, Google Cloud, Amazon Web Services, Deloitte, PwC, EY, KPMG and large law firms.

The work at risk includes paralegal review, document checking, junior legal research, compliance administration, audit support and routine drafting.

The public risk is access and accountability. If law becomes cheaper for corporations but not clearer for citizens, automation isn’t helpful.

TRANSPORT, ENERGY AND INFRASTRUCTURE

This includes traffic systems, railway maintenance, ferry demand, port logistics, airport systems, delivery routing, smart meters, energy grid balancing, wind farm optimisation, water networks, waste collection, building controls and predictive maintenance.

These workloads are controlled by transport agencies, energy companies, utilities, infrastructure firms, councils, regulators and cloud providers. Relevant actors include National Grid, National Energy System Operator, SSEN, ScottishPower, Ofgem, Network Rail, Transport Scotland, ferry operators, water companies, port operators, airports, Siemens, Schneider Electric, IBM, Microsoft and Amazon Web Services.

The work at risk includes monitoring, scheduling, dispatch, routine planning, administrative reporting and some engineering support.

The vulnerability is severe. If the digital systems coordinating energy, transport, water and logistics fail, the failure does not remain online. It appears in dark houses, delayed trains, empty shelves, failed payments, missed deliveries and stalled emergency response.

AGRICULTURE, LAND AND ENVIRONMENT

This includes satellite crop monitoring, livestock systems, soil analysis, forestry modelling, carbon credit verification, peatland monitoring, biodiversity scoring, farm subsidy compliance, drone surveys, water quality prediction, flood modelling, wildfire risk, natural capital accounting and land valuation.

These workloads are controlled by governments, landowners, environmental agencies, universities, agritech companies, carbon market companies, insurers and investors. Key actors include Scottish Government agencies, UK environmental bodies, carbon registry systems, large landowners, natural capital firms, satellite companies, drone companies, Microsoft, Google, Amazon and specialist environmental data firms.

The work at risk includes inspection, monitoring, clerical agricultural administration, mapping, valuation and compliance work.

The rural risk is that the data about land becomes more powerful than the people who live and work on it. Once land is reduced to carbon, biodiversity units, satellite layers and investment models, the farmer can become an inconvenience to the spreadsheet.

PERSONAL LIFE AND DIGITAL IDENTITY

This includes photographs, emails, search histories, smart speakers, home cameras, wearable health data, bank transactions, location trails, social media behaviour, dating profiles, children’s learning data, smart meters, connected cars, home energy systems and online identity verification.

These workloads are generated by individuals but controlled mainly by platforms, device makers, advertisers, cloud providers, banks, app companies and sometimes public authorities. Key companies include Apple, Google, Meta, Amazon, Microsoft, Samsung, TikTok, X, PayPal, Visa, Mastercard, Experian, Equifax and thousands of app companies.

The work at risk is indirect: fewer human staff in shops, banks, call centres, offices, travel agencies, customer service, public counters and administration.

The civic risk is direct. A person who cannot operate digitally may become less able to bank, travel, claim, complain, work, learn, receive care or prove who they are. The digital world is sold as convenience. It becomes compulsion when the non digital route is removed.

AI DATA CENTRES MAY ABSORB OR AUTOMATE THESE WORKLOADS:

Government and public administration: tax queries, tax fraud detection, benefit eligibility, welfare fraud scoring, planning triage, public consultation analysis, licence renewals, passport administration, immigration review, procurement screening, grant scoring, council contact centres, housing list analytics, homelessness forecasting, social care summaries, child protection record search, court transcription, tribunal review, police evidence indexing, prison modelling, parliamentary correspondence, environmental permits, agricultural subsidy administration, public health alerts and emergency planning.

Health and life sciences: medical imaging, radiology triage, pathology analysis, clinical notes, transcription, appointment scheduling, discharge letters, ambulance prediction, bed management, NHS call triage, remote monitoring, wearables, drug discovery, protein modelling, trial matching, genomics, pharmacovigilance, prescription alerts, medical coding, health insurance claims, mental health chatbots, surgical planning, infection modelling, vaccine research and elderly care monitoring.

Education and research: automated tutoring, lesson planning, worksheets, marking support, plagiarism detection, admissions screening, student risk analytics, university chatbots, library search, literature review, grant drafting, academic translation, lecture transcription, curriculum mapping, exam administration, timetabling, online course generation, feedback analysis, research cleaning, lab automation, citation checking, dissertation support, language learning, special needs prompts and parent communications.

Commerce and retail: product recommendation, personalised pricing, stock forecasting, warehouse routing, delivery optimisation, returns processing, complaint chatbots, voice call automation, sales scoring, subscription retention, loyalty analysis, advertising placement, product descriptions, review moderation, market research, demand prediction, checkout fraud, payment disputes, supplier risk, procurement automation, invoice matching, replenishment, workforce scheduling, mystery shopping analytics and sentiment tracking.

Finance, insurance and professional services: credit scoring, mortgage pre screening, loan review, insurance underwriting, claims triage, fraud detection, anti money laundering, know your customer checks, pension administration, investment research, portfolio modelling, audit sampling, tax preparation, expense checking, payroll anomaly detection, due diligence, contract review, legal disclosure, legal research, regulatory scanning, board summaries, compliance monitoring, procurement law, intellectual property search and risk register drafting.

Media and creative economy: news aggregation, headline testing, social drafting, image generation, video editing, audio cleanup, podcast transcription, synthetic voiceover, translation, dubbing, copywriting, advertising concepts, layout design, stock image replacement, book summaries, manuscript assessment, cover mockups, game assets, character dialogue, animation support, music stems, jingle production, press release rewriting, newsletter production and audience analytics.

Software, engineering and industry: code generation, code testing, bug triage, documentation, cyber scanning, DevOps automation, chip simulation, digital twins, building information modelling, engineering drawings, predictive maintenance, factory robotics, quality inspection, supply chain modelling, energy forecasting, grid balancing, wind optimisation, battery management, oil and gas modelling, carbon accounting, construction scheduling, materials discovery, robotics training, industrial safety, water network modelling and waste plant optimisation.

Agriculture, land and environment: crop monitoring, livestock tracking, soil analysis, fertiliser planning, forestry modelling, carbon credit verification, peatland monitoring, biodiversity scoring, habitat mapping, invasive species detection, flood modelling, wildfire risk, fisheries monitoring, aquaculture feeding, farm subsidy compliance, drone surveys, weather risk, food supply forecasting, animal movement records, land valuation, rural planning, environmental impact screening, renewable siting, water quality prediction and natural capital accounting.

Security, defence and policing: satellite analysis, drone video interpretation, maritime tracking, signals intelligence, cyber detection, malware analysis, border risk scoring, biometric matching, facial recognition, licence plate recognition, communications monitoring, battlefield logistics, autonomous vehicle training, target recognition, missile defence simulation, electronic warfare, weapons testing data, military procurement modelling, defence maintenance prediction, disinformation tracking, extremist content detection, crowd monitoring, disaster routing, emergency modelling and infrastructure threat scoring.

Personal and platform data: search histories, email filtering, photo recognition, smart speaker commands, home camera footage, mobile location trails, payment behaviour, streaming recommendations, dating app matching, social media feeds, gaming behaviour, child safety monitoring, wearable fitness data, smart meter data, connected car telemetry, home energy optimisation, personal finance apps, online identity checks, spam filtering, document assistants, calendar scheduling, travel recommendations, food delivery prediction, ride hailing allocation and targeted political messaging.

SCOTLAND’S AI EXPOSURE

No public source can honestly say exactly how many Scottish jobs AI will replace. The better measure is exposure.

The World Economic Forum’s Future of Jobs Report 2025 says employers expect 92 million jobs to be displaced globally by 2030, while 170 million may be created. The Institute for Public Policy Research has warned that up to 7.9 million UK jobs could be displaced by generative AI in a worst case scenario, with a central scenario of 4.4 million. The Tony Blair Institute has estimated that full AI adoption by UK firms could save almost a quarter of private sector workforce time, equivalent to the annual output of six million workers.

Scotland had about 2.645 million people in employment in early 2026, compared with 34.328 million across the UK. Scotland therefore represents roughly 7.7 per cent of UK employment. If UK wide exposure were distributed evenly, a 4.4 million job displacement scenario would imply around 339,000 Scottish jobs exposed. A 7.9 million scenario would imply around 608,000. Those are not predictions. They are scale markers.

The jobs most exposed are not only factory jobs. They are clerical jobs, call centre jobs, junior legal jobs, junior finance jobs, translation jobs, design jobs, writing jobs, coding jobs, public administration jobs, scheduling jobs, claims jobs, monitoring jobs, customer service jobs and the ordinary middle layer of office work that allowed people to earn a living without owning a platform.

THE POWER QUESTION

A hyperscale AI data centre is an electricity demand attached to an automation machine.

If such centres are treated as essential infrastructure, they will compete for grid capacity with homes, transport, heating, manufacturing, hospitals and public services. If society then moves its basic functions into those data centres, it becomes dependent on their continuous operation.

That is the quiet trap. The more services become digital only, the more the citizen must enter the digital system. The more the citizen must enter the digital system, the more power shifts to the organisations that own the platforms, process the data and control access.

A digital society without reliable electricity is not modern. It is helpless.

A society that has removed the bank branch, the paper form, the staffed counter, the local office, the cash option, the printed record and the human telephone line cannot simply return to normal when systems fail. It has destroyed the bridge back.

SOURCES
UK Government, AI Opportunities Action Plan, 13 January 2025

UK Government, Data centres to be designated Critical National Infrastructure, 12 September 2024

UK Government, Data centres: Cyber Security and Resilience Bill factsheet, 6 March 2026

Audit Scotland, Tackling Digital Exclusion, August 2024

Scottish Government, Digital Strategy for Scotland: Vision Statement, 18 November 2025

Financial Conduct Authority, Access to cash rules, PS24/8, 24 July 2024

Financial Conduct Authority, Access to cash consumer guidance, updated 8 May 2026

World Economic Forum, Future of Jobs Report 2025

Institute for Public Policy Research, Up to 8 million UK jobs at risk from AI unless government acts, 27 March 2024

Tony Blair Institute, The Impact of AI on the Labour Market, 8 November 2024

UK Government, Assessment of AI capabilities and the impact on the UK labour market, 28 January 2026

Scottish Government, Labour Market Trends: April 2026