Your best people are already doing two jobs. You are only paying for one.
By Dr Luke Soon Author of Genesis: Human Experience in the Age of Artificial Intelligence & Synthesis
Somewhere in your organisation, right now, one of your best people has quietly taken a second job.
They praised the speed. They banked the gain. They never asked who bore the strain. The ledger smiled. The worker sighed. And somewhere, a first rung quietly died.
Same payroll. Same title. Same band. But sometime in the last eighteen months they became the one who fixes the prompts. The one who checks the model’s maths before it reaches a client. The unofficial translator between the machine and everyone still afraid of it.
Nobody hired them for this. Nobody is paying them for it.
Somewhere in your organisation, right now, one of your best people has quietly taken a second job.
Same payroll. Same title. Same band.
But sometime in the last eighteen months they became the one who fixes the prompts. The one who checks the model’s maths before it reaches a client. The unofficial translator between the machine and everyone still afraid of it.
Nobody hired them for this. Nobody is paying them for it.
You know exactly who I mean.
Welcome to the unpriced labour of the AI era.
The comfortable story the whole industry is telling
We have been handed a reassuring story about AI and work. And, remarkably, every major research house is now telling a version of it.
McKinsey calls the destination “superagency”: humans amplified by AI, unlocking new creativity and productivity. BCG urges companies to stop merely deploying AI and start reshaping workflows. Bain says the winners are those who redesign work rather than automate their “workflow debt”. PwC and the World Economic Forum tell employers to redesign roles and safeguard the pipeline. The CHRO playbooks say: deconstruct the tasks, reskill the people, mind the talent flow.
It is all true.
And every one of these reports describes the same strange new animal without ever naming the thing that matters most.
Because AI did not merely make existing work faster. It created an entire new layer of work. Reviewing. Correcting. Orchestrating. Judging. Teaching. Then it handed the bill to the people least able to refuse it.
The dividend is real. A large part of it is being financed by labour nobody is counting, costing, or paying for.
Let me state the claim plainly, because it is the whole argument:
The early productivity gains of AI are being subsidised by your own people. And the subsidy is invisible, because it never appears in a job description.
Productivity, it turns out, is just overtime with a better publicist.
Five houses, one animal
Here is what makes this more than an opinion. Read the flagship studies side by side, and the same phenomenon surfaces in five different vocabularies.
McKinsey’s Superagency survey found that leaders think 4% of their employees use generative AI for at least 30% of their daily work. The real figure is 13%. More than three times higher. The work is already happening. Leadership simply cannot see it.
BCG’s latest global AI at Work survey, nearly 12,000 workers across 14 markets, found that almost half of employees now spend more time directing and managing AI than doing the task themselves. BCG calls the lived experience a “joy paradox”: AI makes the work better and harder at the same time, stripping out dull tasks while quietly adding new review burdens and new responsibility for choosing wisely.
Bain, tracking software teams, watched “review time” balloon into the dominant bottleneck. An unbudgeted cost that eats the very productivity the tools promised.
Anthropic’s Economic Index, built from millions of real AI conversations, found that most AI use is not automation at all. It is augmentation: humans validating, iterating on, and correcting the model’s output. Read that again. The single most common thing people do with the most advanced AI on earth is check its work.
And the WEF and PwC study of entry-level workers found that 68% feel more productive with AI. While 45% report working longer hours because of it.
WHAT EACH HOUSE CALLS THE SAME UNPAID WORK--------------------------------------------------------McKinsey ....... the AI use leaders can't even see (4% vs 13%)BCG ............ the "joy paradox": better and harder at onceBain ........... "review time" no one put in the budgetAnthropic ...... most AI use is humans checking the AIWEF x PwC ...... 68% more productive, 45% longer hoursPwC Barometer .. "seniorisation" of the entry rungStanford ....... the vanishing 22-to-25-year-old--------------------------------------------------------One phenomenon. Seven names. Zero line items.
Seven names. Zero line items.
That is the tell everyone is trained to walk straight past.
Seven research houses described the same animal. Not one of them brought the invoice.
This new labour shows up at two points in a career. One at the very start. One running right through the middle of your organisation.
Face one: seniorisation, the first rung priced as if it were still the first rung
I have argued before that AI is driving a Great Skills Recomposition. That the pyramid is folding into an hourglass. That the apprenticeship pipeline must be equipped with AI, not amputated by it.
This is the next turn of that argument. And the data has caught up with it hard.
The entry-level job did not disappear. It was quietly promoted. And never repriced.
PwC’s Global AI Jobs Barometer, drawn from over a billion job advertisements, finds that in the most AI-exposed occupations, 52% of the new skills now appearing in entry-level postings are ones that used to belong to experienced professionals: judgement, stakeholder management, strategic decision-making. In the least-exposed occupations, that figure is 7%. PwC calls it seniorisation. Entry roles in exposed fields are now roughly seven times more likely to demand skills that historically arrived a decade later. Redrawn entry roles have grown 35% since 2019. Traditional entry openings shrank 10%.
We are asking twenty-two-year-olds to arrive performing like thirty-five-year-olds. And we pay them, title them, and induct them as though nothing has changed.
We handed the graduate a veteran’s job and a rookie’s wage, and we called it opportunity.
Stanford’s Digital Economy Lab puts a number on the consequence. In Canaries in the Coal Mine, Brynjolfsson, Chandar and Chen tracked millions of ADP payroll records and found a 13% relative decline in employment for workers aged 22 to 25 in the most AI-exposed occupations since late 2022. Around 20% for young software developers. Meanwhile, older and less-exposed workers held steady or grew.
Two details matter here, and they are exactly the kind headline writers skip.
The adjustment came through employment, not pay: firms cut junior hiring rather than junior wages. And the declines cluster precisely where AI automates rather than augments. The authors are careful, too. With their strictest controls the AI-linked timing becomes clearest from 2024, so this is a structural repricing, not one product’s shockwave.
The senior-grade judgement is demanded. It is simply not priced.
Face two: shadow jobs, the unrecognised payroll running through your middle
Now walk up from the entry rung into the body of the organisation. You meet the second face.
When AI is deployed but job architecture is left frozen, people informally become the AI super-user, the prompt-fixer, the one who checks the machine’s output and quietly teaches everyone else. Real new responsibility. For quality. For risk. For adoption itself. And it appears in no job description, no career path, no pay band.
I call these shadow jobs: the invisible second payroll that AI created and nobody authorised.
No title, no band, no line on the sheet, only the hum of a job left incomplete. They mend what it breaks and they check what it writes, they carry the days and they swallow the nights.
Once you have the concept, you see it everywhere in the industry data. McKinsey’s mid-career employees becoming their team’s unofficial AI help desk. BCG’s “half now manage AI more than they do the work”, which BCG frankly calls a managerial revolution happening without managerial titles or pay. Bain’s swelling, unbudgeted review burden. Anthropic’s finding that validation and correction are the dominant mode of human and AI work.
These are not five different findings. They are five sightings of the same unpaid role.
Left alone, shadow jobs do three quiet kinds of damage.
They concentrate critical capability in people you cannot see, and therefore cannot protect. Retention risk wearing a disguise.
They turn adoption into a lottery that depends on whether a team happens to contain a generous super-user. BCG finds frontline adoption has stalled at a “silicon ceiling”, and only a quarter of frontline workers say their leaders give them enough support.
And they push governance, the checking, the ethics, the data-quality control, into the margins of the day. Done off the side of the desk. Or not at all.
Seniorisation is unpriced labour at the start of a career. Shadow work is unpriced labour in the middle of one. Same mechanism. Same missing line on the invoice.
The cheapest hire in your company is the second job your best person is already doing for free.
The productivity paradox is not a paradox. It is a transfer.
Sit with the numbers and the famous “paradox” dissolves.
68% of entry-level workers say AI has made them more productive. 45% say they are working longer hours. BCG finds 42% of employees now save at least a full working day each week, and 66% get little or no guidance on how to reinvest that reclaimed time, so it silently refills with more work. Bain finds that basic AI assistants deliver perhaps 10 to 15% productivity, most of which never reaches the P&L at all.
There is no paradox here. There is a transfer.
THE AI PRODUCTIVITY INVOICE (per knowledge worker)---------------------------------------------------BOOKED BY THE FIRM Productivity uplift ............ recognised, bankedPAID BY THE WORKER (unbilled) Longer hours ................... 45% report more time Reclaimed time, re-spent ....... 66% given no plan for it Senior judgement, junior grade seniorisation Prompt-fixing, output-checking shadow jobs Skill churn / relevance fear ... 28% doubt their skills---------------------------------------------------NET: the dividend is real. The bill is just sent to the wrong desk.
The productivity lands on the firm’s side of the ledger, booked as efficiency, margin, headcount avoided. The cost lands on the worker’s side: longer hours, absorbed judgement, unrecognised responsibility, low-grade relevance anxiety. And it is booked nowhere at all.
A saved hour that nobody chose to spend is simply unpaid overtime in a smarter suit.
Why the ledger broke: AI acts on the task, not the role
None of this is malice. It is an accounting failure with a very old root.
Two decades ago, David Autor and Daron Acemoglu reframed the economics of work. A job is not an indivisible thing. It is a bundle of tasks.
AI understood this before our HR systems did. It does not replace roles wholesale. It reaches inside them and rearranges the tasks: automating some, co-piloting others, leaving a hard human core of relationships, context and ethical judgement. McKinsey’s Global Institute now estimates that more than half of current work hours are technically automatable. Anthropic’s data shows AI already touching a meaningful share of tasks across roughly half the jobs it studied.
But our job architecture, the descriptions, the skills taxonomies, the pay bands, the promotion criteria, is still written at the level of the role. So when AI rewrote the tasks, the ledger that prices the work never updated.
That is the entire crisis in one line. AI repriced the work, and we forgot to reprice the pay.
The atomic unit of work changed. The unit of account did not. The result is an economy running a second payroll it refuses to print.
The uncomfortable part: the fix already pays, and we are skipping it anyway
Here is where the industry’s own evidence turns from diagnosis into indictment.
The firms that actually redesign work around AI, that reprice the tasks and reshape the roles, are not sacrificing performance. They are winning. The WEF and PwC find that organisations redesigning work alongside AI adoption are twice as likely to post strong financial performance. Bain finds that companies linking workforce and workflow deliberately deliver more than twice the total shareholder return of their peers, and that full lifecycle redesign yields 25 to 30% productivity against 10 to 15% for bolt-on tools. BCG finds its “Reshape” companies capture the value that “Deploy” companies never see.
And yet.
McKinsey reports that only 1% of companies consider themselves mature in AI deployment. Bain finds just 23% can tie their AI initiatives to real revenue or cost. Almost everyone is banking the productivity story. Almost no one is doing the repricing that the same research says pays double.
Let me say it plainly.
We are not failing to price the new labour because it is unprofitable to price it. We are failing because unpaid labour is, for a few quarters, cheaper than paid labour. That is the uncomfortable sentence sitting under all the optimism.
Unpaid labour is cheaper than paid labour, for exactly as long as your best people are still in the building.
The Asia-Pacific and Global South sharpening
Nowhere is this more urgent than in the emerging economies now leading the world in AI use.
In Eastern Asia, roughly three in four young workers, 75%, sit in roles with medium-to-high AI exposure, against a global average of 37%. BCG finds the Global South out-adopting the West outright: India at 92% regular use, the Middle East at 87%, with the United States near the bottom of the table. And those same high-adoption regions report the highest fear of job loss.
The unpriced-labour problem is not evenly spread. It is concentrated precisely where our youngest, most digitally fluent, most economically mobile workers are. And where the social safety net is thinnest.
This is why Singapore’s instinct matters well beyond Singapore. SkillsFuture. WSG’s job-redesign support. The national push on AI capability. These are, in effect, public attempts to put a price on the new labour before the market can hide it.
Not every economy is moving this early. The ones that do not will discover the deficit later, and pay it with interest.
The HX cost: EX debt always matures into CX debt
Regular readers know my core equation. HX = CX + EX. Human Experience is the sum of Customer Experience and Employee Experience, and neither survives long without the other.
Unpriced labour, in my HX framework, is EX debt. And EX debt never stays put.
The over-stretched super-user burns out and leaves, taking your real AI capability out the door. The seniorised graduate, promised growth and handed grade-inflated pressure, disengages. BCG’s “joy paradox” is EX debt in real time: satisfaction and cognitive load rising together, the honeymoon fading inside a year unless leaders intervene.
And depleted, disengaged people do not create the moments that generate loyalty, trust and Return on Experience (ROX). EX debt matures, reliably, into CX debt.
You cannot extract your way to a superior customer experience on the backs of people running two jobs for one wage. The maths only looks like it works. For two or three quarters. Right up until it doesn’t.
Every unpaid hour is a small loan from your workforce. Attrition is simply the day it is called in.
What leaders should actually do
The playbooks are precise about the mechanics and silent about the money. So let me add the missing lines.
- Price the new work. Before you book a single point of AI productivity, run a task-level pass on your critical roles, automated, human-and-AI, human-only, and ask the question every deck skips. Who is now doing new work, and are we paying them for it?
- Codify the shadow jobs. The super-user, the prompt engineer, the AI quality lead, the workflow designer. Pull them out of the shadows and into the job catalogue, with titles, progression and pay. What you recognise, you can retain. What you hide, you lose. Usually to a competitor who noticed first.
- Reprice the first rung. If entry roles now demand senior judgement, teach it deliberately and reward it honestly. Do not smuggle senior expectations in under an unchanged band. Redesign the rung. Do not merely raise the bar and hope.
- Budget the reclaimed time. If AI hands a worker back a day a week, decide, with them, where that day goes. Unplanned, it refills with more work and the dividend evaporates. Reinvested deliberately, it becomes learning, judgement, or genuine capacity.
- Make the ledger visible. Put the unpriced labour on the balance sheet as a real cost of adoption. A transformation that quietly runs on unpaid intensity is not a transformation. It is a loan against your workforce, and the repayment is attrition.
- Govern in the open. Fold responsible-use, data-quality and ethical-oversight duties into role descriptions and pay, not into side-of-desk goodwill. Trusted agents need trusted, resourced humans around them.
None of this slows AI down. The whole weight of the industry’s own evidence says the opposite. Pricing the labour is not a tax on the dividend. It is how the winners earn it.
The Fork
I have long argued that we stand at a Fork. A narrow window of civilisational choices that decides whether AI delivers Short-term Turbulence for Long-term Abundance, or something meaner. The Star Trek future, or the Mad Max one.
Here is that Fork, rendered in the smallest, most human unit imaginable. A single payslip.
Down one path, we treat AI’s productivity as free money. We book the dividend, bury the cost, let the shadow jobs multiply and the first rung drift out of reach. It works beautifully. Until the super-users quit, the graduates give up, the pipeline starves, and the customer feels the hollowing a few quarters later.
Down the other, we do the unglamorous thing the data already rewards. We reprice the work. We name the shadow jobs. We rebuild the rung. We put the invisible labour back on the ledger where it belongs, and we share the surplus AI creates with the people creating it.
Every major house has now mapped the same terrain. McKinsey sees the hidden usage. BCG names the paradox. Bain counts the review burden. Stanford watches the canaries. What none of them will do is sign the cheque.
That part is leadership. And it is on the table right now: in the next performance cycle, in the next AI business case, in the next job description you sign off without reading twice.
Two futures fork from an ordinary wage: one hollows the ladder, one turns the page. The spreadsheet stays silent. The choosing is our own. Build it with people, or build it alone.
AI created a new kind of work. The only question left is whether we have the honesty to pay for it.
Build the future with your people. Not to them.
Dr Luke Soon Genesis: Human Experience in the Age of Artificial Intelligence | Synthesis: The Superintelligence Protocol Partner, PwC | Singapore
Sources referenced: WEF and PwC, Artificial Intelligence and the Future of Entry-Level Work (2026); PwC Global AI Jobs Barometer (2026); McKinsey, Superagency in the Workplace, and MGI, The Rise of the Human and AI Workforce (2025 to 2026); BCG, AI at Work (2025 to 2026); Bain Generative AI Survey and Technology Report (2025); Stanford Digital Economy Lab and HAI, Brynjolfsson, Chandar and Chen, Canaries in the Coal Mine? (2025); Anthropic Economic Index (2025 to 2026); Autor and Acemoglu, task-based framework.
#FutureOfWork #ArtificialIntelligence #HumanExperience #Seniorisation #ShadowWork #WorkforceTransition #AgenticAI
Welcome to the unpriced labour of the AI era.
Te part of it is being financed by labour nobody is counting, costing, or paying for.
Let me put the claim plainly, because it is the whole argument:
The early productivity gains of AI are being subsidised by your own people – and the subsidy is invisible because it never appears in a job description.
There are two faces to this. One sits at the very start of a career. One runs right through the middle of your organisation.
Phase one: seniorisation – the first rung, priced as if it were still the first rung
I have argued before that AI is driving a Great Skills Recomposition — that the pyramid is folding into an hourglass, and that the apprenticeship pipeline must be equipped with AI, not amputated by it. This is the next turn of that argument.
The entry-level job did not disappear. It was quietly promoted – and never repriced.
In the most AI-exposed occupations, 52% of the new skills now appearing in entry-level job postings are skills that used to belong to experienced professionals: judgment, stakeholder management, strategic decision-making. In the least-exposed occupations, that figure is 7%. PwC’s Barometer calls this seniorization. Entry-level roles in highly exposed fields are now roughly seven times more likely to demand skills that historically arrived a decade later.
We are asking twenty-two-year-olds to arrive performing like thirty-five-year-olds — while paying them, titling them, and inducting them as though nothing has changed.
The market has already voted. Redrawn entry roles have grown 35% since 2019; traditional entry openings shrank 10%. And vacancies in the most AI-exposed quartile are the only group to have flatlined since 2012, while every other quartile kept climbing.
Before anyone builds a doom headline: the slowdown began nearly a year before ChatGPT shipped. This is a structural repricing, not one product’s shockwave – Brynjolfsson, Chandar and Chen find the same entry-level squeeze in the US data.
The senior-grade judgment is demanded. It is simply not priced.
Phase two: shadow jobs –> the second payroll running through your middle
Now walk up from the entry rung into the body of the organisation, and you meet the second face.
PwC’s own CHRO blueprint names it with unusual honesty: when AI is deployed but job architecture is left frozen, people informally become the AI super-user, the prompt-fixer, the one who checks the machine’s output. Real new responsibility – for quality, for risk, for teaching everyone else – that appears in no job description, no career path, and no pay band.
I call these shadow jobs: the invisible second payroll that AI created and nobody authorised.
Left alone, shadow jobs do three quiet kinds of damage. They concentrate critical capability in people you cannot see and therefore cannot protect — that is retention risk wearing a disguise. They turn adoption into a lottery that depends on whether a team happens to contain a generous super-user — that is inconsistency. And they push governance, the checking and the ethics and the data-quality control, into the margins of people’s days, where it is done off the side of the desk or not at all – that is risk.
Seniorisation is unpriced labour at the start of a career. Shadow work is unpriced labour in the middle of one. Same mechanism. Same missing line on the invoice.
The productivity paradox is not a paradox. It is a transfer.
Here is the number everyone quotes and nobody sits with: 68% of entry-level workers say AI has made them more productive – and 45% say they are working longer hours because of it. A further 28% believe half or fewer of their current skills will still be relevant in three years.
More productive with AI ████████████████░░░░ 68%Working longer hours ███████████░░░░░░░░░ 45%Doubt half their skills ███████░░░░░░░░░░░░░ 28%
Read those together and the “paradox” dissolves. There is no paradox. There is a transfer.
The productivity lands on the firm’s side of the ledger — booked as efficiency, as margin, as headcount avoided. The cost lands on the worker’s side — as longer hours, absorbed judgment, unrecognised responsibility, low-grade relevance anxiety — and is booked nowhere at all.
THE AI PRODUCTIVITY INVOICE (per knowledge worker)---------------------------------------------------BOOKED BY THE FIRM Productivity uplift ............ recognised, bankedPAID BY THE WORKER (unbilled) Longer hours ................... 45% report more time Senior judgment, junior grade . seniorization Prompt-fixing, output-checking shadow jobs Skill churn / relevance fear ... 28% doubt their skills---------------------------------------------------NET: the dividend is real. The bill is just sent to the wrong desk.
The dividend is genuine. It is simply being financed by people who never agreed to lend.
Why the ledger broke: AI acts on the task, not the role
None of this is malice. It is an accounting failure with a very old root.
David Autor spent two decades reminding economists that work is not made of jobs; it is made of tasks. AI understood this before our HR systems did. It does not replace roles wholesale — it reaches inside them and rearranges the tasks: automating some, co-piloting others, leaving a hard human core of relationships, context and ethical judgment that no model can replicate.
But our job architecture — the descriptions, the skills taxonomies, the pay bands, the promotion criteria — is still written at the level of the role. So when AI rewrote the tasks, the ledger that prices the work never updated.
That is the entire crisis in one line: AI repriced the work, and we forgot to reprice the pay. The atomic unit of work changed. The unit of account did not.
The Asia-Pacific sharpening
Nowhere is this more urgent than in Asia-Pacific.
In Eastern Asia, roughly three in four young workers – 75% – sit in roles with medium-to-high AI exposure, against a global average of 37%. The unpriced-labour problem is not evenly spread. It is concentrated precisely where our youngest, most digitally fluent, most economically mobile workers are.
This is why Singapore’s instinct matters well beyond Singapore. SkillsFuture, WSG’s job-redesign support, the national push on AI capability — these are, in effect, public attempts to put a price on the new labour before the market can hide it. Not every economy is moving this early. The ones that do not will discover the deficit later, and pay it with interest.
The HX cost: EX debt always matures into CX debt
Regular readers know my core equation: HX = CX + EX. Human Experience is the sum of Customer Experience and Employee Experience, and neither survives long without the other.
Unpriced labour, in this language, is EX debt – and EX debt never stays put. The over-stretched super-user burns out and leaves, taking your real AI capability out the door with them. The seniorised graduate, promised growth and handed grade-inflated pressure, disengages. And depleted, disengaged people do not create the moments that generate loyalty, trust and Return on Experience (ROX).
You cannot extract your way to a superior customer experience on the backs of people running two jobs for one wage. The maths does not work. It only looks like it works — for two or three quarters, right up until it doesn’t.
What leaders should actually do
The corporate playbooks are precise about the mechanics and silent about the money. So let me add the missing lines.
- Price the new work. Before you book a single point of AI productivity, run a task-level pass on your critical roles — AI-only, human-and-AI, human-only — and ask the question the decks skip: who is now doing new work, and are we paying them for it?
- Codify the shadow jobs. The super-user, the prompt engineer, the AI quality lead, the workflow designer — pull them out of the shadows and into the job catalogue, with titles, progression and pay. What you recognise, you can retain. What you hide, you lose.
- Reprice the first rung. If entry roles now demand senior judgment, teach it deliberately and reward it honestly. Do not smuggle senior expectations in under an unchanged band. Redesign the rung; don’t merely raise the bar and hope.
- Make the ledger visible. Put the unpriced labour on the balance sheet as a real cost of adoption. A transformation that quietly runs on unpaid intensity is not a transformation — it is a loan against your workforce, and the repayment is attrition.
- Govern in the open. Fold responsible-use, data-quality and ethical-oversight duties into role descriptions and pay — not into side-of-desk goodwill. Trusted agents need trusted, resourced humans around them.
None of this slows AI down. The evidence cuts the other way. Organisations that redesign work alongside AI adoption are twice as likely to post strong financial performance as those that merely deploy the tools. Dropbox expanded rather than cut its graduate intake and saw promotion rates improve within two to three years. Pricing the labour is not a tax on the dividend. It is how you keep earning it.
The Fork
I have long argued that we stand at a Fork — a narrow window of civilisational choices that decides whether AI delivers Short-term Turbulence for Long-term Abundance, or something meaner.
Here is that Fork, rendered in the smallest, most human unit imaginable: a single payslip.
Down one path, we treat AI’s productivity as free money. We book the dividend, bury the cost, let the shadow jobs multiply and the first rung drift out of reach. It works beautifully — until the super-users quit, the graduates give up, the pipeline starves, and the customer feels the hollowing a few quarters later.
Down the other, we do the unglamorous thing. We reprice the work. We name the shadow jobs. We rebuild the rung. We put the invisible labour back on the ledger where it belongs, and we share the surplus AI creates with the people creating it.
The technology will not choose for us. The spreadsheet will not choose for us. This is a leadership choice, and it is on the table right now – in the next performance cycle, in the next AI business case, in the next job description you sign off without reading twice.
AI created a new kind of work. The only question left is whether we have the honesty to pay for it.
Build the future with your people – not to them.
Dr Luke Soon Genesis: Human Experience in the Age of Artificial Intelligence | Synthesis: The Superintelligence Protocol
#FutureOfWork #ArtificialIntelligence #HumanExperience #Seniorization #ShadowWork #WorkforceTransition #AgenticAI
Genesis: Human Experience in the Age of Artificial Intelligence | Synthesis: The Superintelligence Protocol 
Leave a comment