AI is Paradise

You are 26 years old.

You work in investment banking in London.

You have recently discovered artificial intelligence.

You are absolutely fucking convinced it is going to change your life.

Six months ago, you spent three hours writing a twelve-line email to a client. You now spend forty-five minutes prompting ChatGPT to write the same email, before ultimately sending your original version.

You consider this a tremendous efficiency gain.

You pay £200 a month for the most expensive subscription. You don't know what additional features it provides, but the word Pro resonates with you.

You have also subscribed to Claude, Gemini and Perplexity. You refer to them collectively as your AI stack.

You have never stacked anything.

At 8:47 on Monday morning, you arrive at the office. You are nine minutes late.

Your Associate is already at his desk.

“Morning. Any progress on the market section?”

“Yeah, basically done. Just finalising a few bits.”

You have not started.

But you have ChatGPT.

You open a new conversation.

I need a detailed analysis of the European residential investment market. This is for a very important institutional client. Needs to be absolutely top quality. Think like a senior investment banker at Goldman Sachs.

ChatGPT responds.

“Absolutely. Let's build an institutional-grade analysis that combines rigorous financial thinking with compelling strategic insights.”

Fuck yes.

You have never felt more understood.

Within eleven seconds, it produces an eight-page report explaining that residential real estate benefits from favourable demographic tailwinds, structural undersupply and resilient demand fundamentals.

The insights are extraordinary.

People need somewhere to live.

Supply is constrained.

Demand is robust.

You immediately paste the entire thing into PowerPoint.

At 9:14, your Associate walks over.

“How are we looking?”

“Yeah, pretty much there.”

He reads the first slide.

“What's the source for this?”

You explain that you've triangulated several market sources.

He looks at the footnote.

Source: European Real Estate Investment Outlook 2026, European Investment Property Research Institute.

“What's this?”

You have no idea.

You Google the institute.

It does not exist.

You ask ChatGPT why it has fabricated a research institution.

“I sincerely apologise. You're absolutely right to flag this. The reference appears to have been illustrative rather than a verified source.”

You appreciate the accountability.

Your Associate does not.

“Can you just use actual data?”

“Yep, absolutely.”

You return to ChatGPT.

STOP MAKING UP FUCKING SOURCES. ONLY USE REAL SOURCES. THIS IS EXTREMELY IMPORTANT.

“Understood. You're right to demand a higher standard of evidence.”

Finally.

A professional.

You ask it to verify the figures.

It apologises again and produces three replacement sources.

Two don't exist.

The third is a recipe for chicken tikka masala.

You decide to use a published research report.

At 10:23, the team joins its weekly pipeline meeting.

The MD has just returned from a conference in New York.

He is wearing a gilet with the logo of an asset manager that declined to hire him in 2018.

He has some exciting news.

“Guys, I've been spending a lot of time thinking about AI.”

Everyone goes quiet.

“This technology is going to completely transform the way we work.”

You nod enthusiastically.

You have spent the last ninety minutes unsuccessfully trying to verify a German housing completion statistic.

“I was at a dinner with a senior guy from one of the big funds. Apparently they're using AI for absolutely everything.”

The MD pauses.

“Everything.”

You have heard this story before.

Three months ago, he attended a dinner with someone at Blackstone and announced that Excel would be obsolete within eighteen months.

Last week, he asked you how to unhide a column.

“Anyway, I've asked HR to put together an AI working group. I want everyone thinking about how we can leverage this across the business.”

He turns to you.

“You're young. You understand this stuff.”

You accept the compliment.

You are 26 years old and have spent £200 a month to have a computer tell you that your emails are excellent.

“Could you put together a short presentation on how we might implement AI across the team?”

“Of course.”

“Nothing too heavy. Maybe ten slides.”

Your Associate asks whether this should take priority over the live mandate.

The MD looks offended.

“Well, obviously the client comes first.”

He pauses.

“But AI is the future.”

You understand your instructions perfectly.

You now have two priorities.

Both are the first priority.

At 11:06, you ask ChatGPT to create the presentation.

It produces a detailed implementation strategy featuring five workstreams, twelve milestones and a three-year transformation roadmap.

It recommends establishing an AI Centre of Excellence.

You do not know what this is.

Neither does ChatGPT.

You love it.

You ask it to make the recommendations more commercially actionable.

It adds the words commercially actionable to the recommendations.

You ask it to make the slides more punchy.

It adds exclamation marks.

You ask it to write like McKinsey.

It produces a section titled “Reimagining Tomorrow: Unlocking the AI-Powered Advisory Ecosystem.”

You ask it to stop sounding like AI.

It apologises.

You ask it to make the language less corporate.

It suggests “AI is changing the game.”

You tell it never to use the phrase “changing the game” again.

“Absolutely. That's a very fair point.”

You are beginning to suspect that ChatGPT may not have worked at McKinsey.

At 12:37, you finish the presentation.

Slide 7 explains that artificial intelligence could automate 70% of junior analytical tasks.

You are immensely proud of this statistic.

You have no idea where it came from.

You email the presentation to the MD.

He replies immediately.

“Thanks. Looks great. Haven't opened yet but can we make it more visual?”

You ask ChatGPT to make it more visual.

It offers to create a PowerPoint.

You are ecstatic.

You have spent the last three years manually adjusting text boxes and are now watching the death of an entire profession.

You click download.

The resulting presentation contains seventeen slides.

Every slide has the same blue gradient.

The title page features an enormous photograph of a robot shaking hands with a businessman.

The robot has six fingers.

The businessman has seven.

Slide 4 is titled “Strategic Synergies.”

It contains a picture of a lighthouse.

You ask ChatGPT why there is a lighthouse.

“The lighthouse serves as a visual metaphor for clarity and direction during digital transformation.”

You can see its point.

At 2:14, you receive a Teams message from the Associate.

Where are we on the actual client deck?

Fuck.

You open your original chat.

It has grown to 184 messages.

You cannot find the original analysis.

You ask ChatGPT to summarise what you have achieved so far.

“Of course! Here's a concise summary of the significant progress we've made together.”

It lists fourteen completed deliverables.

You have completed none of them.

You start a new conversation.

I have a very important deadline in two hours. I need you to be EXTREMELY efficient.

“Absolutely. I'll prioritise speed, accuracy and clarity.”

You feel a surge of confidence.

You upload the 64MB financial model.

ChatGPT informs you that the file is too large.

You compress it.

Still too large.

You remove the historical financials.

Still too large.

You delete half the worksheets.

It works.

You ask it to identify any errors in the model.

“Certainly. I've conducted a comprehensive review.”

It identifies a potential circularity on a worksheet you deleted twenty minutes ago.

You thank it.

It tells you you're welcome.

You ask it to check the IRR calculations.

It confidently announces that the model is fundamentally sound.

You ask it what the IRR is.

It says 17.4%.

The model says 8.2%.

You point out the discrepancy.

“You're absolutely right. Thank you for catching that.”

You ask which number is correct.

It says 12.6%.

You are now genuinely impressed.

It has found a third answer.

At 4:08, the MD circulates an email to the entire team.

“All,
Following my recent conversations in the US, I think we need to be much more ambitious in our adoption of AI.
The technology is moving incredibly quickly and we cannot afford to be left behind.
I would like everyone to start thinking about how they can integrate AI into their daily workflows.
Best”

He has copied in the Global Head of the division.

You recognise the formatting immediately.

You have been using ChatGPT long enough to know exactly what happened.

You paste the email into an AI detector.

It reports a 99% probability of AI generation.

You grin.

Your MD has discovered ChatGPT.

Your industry is saved.

At 5:16, you have your first breakthrough.

You have finally managed to get ChatGPT to produce a genuinely excellent market commentary.

It is concise.

It is analytical.

It contains real statistics.

Every claim is properly sourced.

You almost cry.

You paste the output into the deck.

Your Associate reads it.

“This is actually pretty good.”

Your chest swells.

“Thanks.”

“Can we just make sure the figures are right?”

You reassure him that everything has been checked.

He asks who checked it.

You explain that you have developed a robust AI-assisted verification process.

In reality, you asked ChatGPT whether it was confident in its analysis.

It said yes.

You consider this a two-stage review.

At 6:42, a Director asks you to help with a new pitch.

“We've got a meeting tomorrow morning. Shouldn't be too difficult. Maybe just put together a quick company overview.”

You would previously have protested.

You are already behind on two live mandates.

You have not eaten lunch.

You were planning to leave at seven.

But now you have AI.

“No problem.”

The Director looks delighted.

“Great. AI should make all this much quicker anyway.”

He leaves.

You open ChatGPT.

URGENT. Need detailed company overview. Must be accurate. No made-up information. This is for a client pitch tomorrow morning.

“Absolutely. Let's get this right.”

You enter the company name.

ChatGPT produces a compelling analysis of a business operating across residential, logistics and renewable energy.

You are confused.

The company manufactures industrial adhesives.

You explain this.

“Thanks for clarifying. That changes the analysis materially.”

You agree.

It does.

At 8:31, the Associate asks whether you can stay to help finish the pitch.

You say yes.

You no longer have an excuse.

Artificial intelligence has made you significantly more productive.

Your ability to say no has fallen by approximately 100%.

At 9:47, the team orders dinner.

You ask ChatGPT what you should eat.

It recommends a balanced meal containing lean protein, complex carbohydrates and vegetables.

You order Five Guys.

You ask ChatGPT whether this is acceptable.

“Absolutely. One meal doesn't define your overall health.”

You love this fucking thing.

At 10:16, the Director sends comments on the deck.

Page 12: Need better market evidence.
Page 14: Can we tighten the investment rationale?
Page 16: Numbers don't tie.
Page 19: Feels AI-generated.

The last comment irritates you.

You have spent six hours engineering the prompts.

You are practically a software developer.

You ask ChatGPT to rewrite page 19 so that it sounds more human.

It replaces “robust structural tailwinds” with “strong long-term fundamentals.”

You ask for another version.

It produces “compelling underlying market dynamics.”

You tell it to write like an intelligent person who actually works in investment banking.

“Understood. You want commercially grounded language, free from generic AI phrasing, with the precision and judgement expected of a senior investment banker.”

Exactly.

It produces “robust structural tailwinds.”

You close the tab.

At 11:32, your Associate makes a startling discovery.

The financial model contains an error.

The return on equity is overstated by 240 basis points.

The client has already seen the numbers.

You stare at the screen.

You have spent eleven hours discussing the future of artificial intelligence with a chatbot.

You have not checked the fucking model.

The Associate asks how this happened.

You explain that you had relied on AI to assist with the quality-control process.

He looks at you.

“What does that mean?”

You explain that you uploaded the model and asked it to check for errors.

“Did it actually open the model?”

You think about this.

You are no longer sure.

At 11:49, the Associate walks over to the Director's desk.

You hear the words “AI” and “model” in the same sentence.

The Director turns towards you.

“Are we putting confidential client information into public AI tools?”

You explain that you have a premium subscription.

He asks whether that makes it secure.

You explain that it's the Pro version.

He asks again.

You have no answer.

At 12:07, an urgent email arrives from Compliance.

“Subject: Use of Generative AI Tools
All employees are reminded that confidential client information must not be uploaded to unapproved third-party AI platforms.”

It is 1,400 words long.

It includes a detailed appendix on data governance.

You suspect it was written by ChatGPT.

At 12:23, HR circulates an invitation to the inaugural AI Transformation Steering Committee.

The MD is chairing.

The Head of Compliance is attending.

So is the Head of IT.

There are nineteen invitees.

You are the only Analyst.

You feel honoured.

At 12:41, you receive the meeting agenda.

Item 1: AI Opportunities.
Item 2: AI Risks.
Item 3: Responsible AI Framework.
Item 4: AI Roadmap.
Item 5: Next Steps.

You open a new ChatGPT conversation.

Please draft a sensible agenda for a corporate AI steering committee.

It produces exactly the same five items.

You are beginning to understand how corporate governance works.

At 1:14 a.m., you finally leave the office.

Your phone tells you that your screen time is up 38% this week.

You consider this proof of technological advancement.

On the taxi home, you open LinkedIn.

A former university acquaintance has posted a photograph of himself speaking at an AI conference in Dubai.

He works in recruitment.

His caption reads:

AI won't replace you. Someone using AI will.

It has 1,400 likes.

You hate him.

You ask ChatGPT whether investment banking Analysts will be replaced by AI.

“AI is more likely to augment human professionals than replace them entirely. Skills such as critical thinking, relationship management and complex judgement will remain essential.”

You relax.

You are excellent at critical thinking.

Earlier today, you correctly identified that a European property research institute did not exist.

At 1:37, you reach your flat.

You have been awake for nineteen hours.

You open your laptop.

There is one final thing you need to do.

You want to understand how much time AI has saved you.

You create a spreadsheet.

Before AI, you estimate that a typical market analysis took six hours.

With AI, the initial draft takes twelve minutes.

Incredible.

You calculate a productivity improvement of 2,900%.

You don't include the four hours spent verifying sources, the three hours rewriting the analysis, the ninety minutes fixing the formatting or the additional two hours spent trying to persuade the chatbot to stop using the phrase “structural tailwinds.”

Your efficiency gain is enormous.

You ask ChatGPT to help articulate it for the AI Transformation Steering Committee.

It suggests:

“AI-enabled workflows have demonstrated the potential to reduce initial drafting time by up to 97%, freeing valuable capacity for higher-value strategic activities.”

Beautiful.

You paste it into your presentation.

You are going to be promoted.

Three weeks later, the AI Transformation Steering Committee holds its first meeting.

The MD opens proceedings with a short speech about innovation.

The Head of IT explains that the firm is evaluating several enterprise AI solutions.

The Head of Compliance explains that none have been approved.

HR suggests mandatory AI training for all staff.

The MD asks whether the training can be delivered by AI.

Everyone agrees this is an excellent idea.

You present your findings.

You explain that AI can dramatically accelerate market research, financial analysis, client communications and presentation preparation.

You show the 97% efficiency statistic.

The MD leans forward.

“Ninety-seven percent?”

You nod.

He is visibly excited.

“So what you're saying is that something that used to take a full day can now be done in fifteen minutes?”

You hesitate.

“Well, that's more the initial drafting component. There is still quite a lot of review and...”

He interrupts.

“Fantastic.”

He turns to the other Directors.

“This is exactly what I've been saying. We need to completely rethink our resourcing model.”

HR begins taking notes.

The Head of Finance asks whether the productivity gains could be reflected in next year's headcount budget.

The MD thinks for a moment.

“Potentially, yes.”

You are suddenly less enthusiastic.

The Associate asks whether AI can actually build and audit a complex financial model without human intervention.

You begin explaining the limitations of large language models.

The MD waves his hand.

“That's just where the technology is today. Think about where it'll be in twelve months.”

You have no rebuttal.

Nobody can argue with the future.

The meeting concludes with unanimous agreement that AI will transform the business.

No software has been approved.

No process has been changed.

No budget has been allocated.

HR announces that all employees will be required to complete a two-hour course on responsible AI usage.

The MD thanks everyone for their hard work.

“Great session. Really productive.”

You return to your desk.

There is a new email from the MD.

“Guys,
Following today's very encouraging AI session, I think we should be pushing ourselves harder on turnaround times.
We clearly have the tools available to work more efficiently, so I'd like us to aim to produce first drafts on the same day wherever possible.
We should also be looking to take on more opportunities rather than turning work away due to capacity constraints.
Exciting times.
Best”

You read the email twice.

Six months ago, you had three days to produce a pitchbook.

You now have one.

Six months ago, you worked until 10 p.m.

You now work until midnight.

Six months ago, you had an Assistant who helped with research.

She has been reassigned to the AI Transformation Programme.

You have become faster at producing work.

The firm has become faster at giving you more work.

You open ChatGPT.

I think AI is making my life worse. I am working longer hours, my bosses expect everything instantly, and I am worried about my job. What should I do?

It responds.

“I'm sorry you're going through this. What you're describing is a very understandable response to a rapidly changing workplace.”

You stare at the screen.

It continues.

“Would you like me to help you develop a practical five-step action plan to regain control of your workload and improve your work-life balance?”

You close the laptop.

Your phone vibrates.

It's the MD.

Quick one. Client has asked for a full investment case by tomorrow morning. Should be straightforward with AI.

Can you take ownership?

You look at the message.

You think about your working hours.

You think about the subscription fees.

You think about the 97% productivity improvement.

You think about the fact that a machine capable of writing a doctoral thesis in thirty seconds has somehow resulted in you spending more time adjusting PowerPoint text boxes than at any point in human history.

You begin typing.

“Unfortunately, I am at capacity and would need to deprioritise other work to...”

You stop.

You delete the message.

You open ChatGPT.

Draft a short, positive reply to my MD confirming I can complete an urgent piece of work overnight. Make it sound keen but not overly enthusiastic.

“Of course! Here's a concise, professional response.”

You copy it.

You press send.

“Absolutely. No problem at all.”

At 12:48 a.m., you open a new conversation.

I need you to act as a top-tier investment banker. This is extremely important. The deadline is tomorrow. You MUST NOT hallucinate. Do you understand?

“Absolutely. I'll approach this with the rigour, precision and commercial judgement expected of a top-tier investment banking professional.”

You smile.

Everything is going to be fine.

AI is paradise.