MANIFESTO 2026-2028

AI Manifesto

In 24 months the world will be unrecognizable. 56 predictions. Accuracy tracker. A guide to action.

AI

Vladimir Doretsky · vibemarketolog.ru ·
Telegram VK

Predictions Dashboard

56 predictions from mid-2025 to 2028 with accuracy tracker

56
predictions tracked
50%
white-collar jobs at risk (1-5 yrs)
10x
productivity boost with AI tools
2027
full AI autonomy expected
$1T
global AI investments in 2026
$20
/month = best AI in the world
56
Total predictions
0
Evaluated
0%
Average accuracy

AI Revolution Timeline

From first agents to superintelligence

Mid 2025
First commercial AI agents
3 predictions · 85% Accuracy
Late 2025
Mega data centers + AI self-improvement
4 predictions · 82% Accuracy
Early 2026
Coding revolution + open models catch up
4 predictions · 86% Accuracy
Mid 2026
Geopolitics + AI in every business
5 predictions · 68% Accuracy
Late 2026
Market growth + first job losses
8 predictions · 62% Accuracy
Jan 2027
Agent-2: continuous learning
3 predictions · Expected
Feb 2027
Cyber threats + weight theft
2 predictions · Expected
Mar 2027
Breakthroughs + Agent-3
2 predictions · Expected
Apr-Jun 2027
Alignment crisis + AI tutors
6 predictions · Expected
Jul-Aug 2027
Agent-3-mini + protests
4 predictions · Expected
Sep-Oct 2027
Agent-4 + leaks
5 predictions · Expected
2028
Superintelligence era
10 predictions · Expected
01 / Analogy

Remember February 2020

If you followed the news closely, you might have noticed a few people talking about a virus spreading overseas. But most of us weren't paying attention. The stock market was up, kids were in school, you went to restaurants, shook hands, and planned vacations.

Then, over about three weeks, the entire world changed. Your office closed, kids came home, and life rearranged into something you wouldn't have believed if described to you a month earlier.

I think we're in the "this feels overhyped" phase of something much, much bigger than COVID.— Matt Shumer, AI startup founder, 6 years in the industry

I, Vladimir Doretsky, creator of vibemarketolog.ru, completely share this view. Over the past months I've been witnessing something incredible: AI that just six months ago was a "useful helper" has become a full-fledged specialist. It doesn't just help — it does the work better than most professionals.

I'm writing this not to scare you. I'm writing because I believe that the biggest advantage you can get right now is simply being earlier than everyone else.

Start using AI now
02 / Personal Experience

I Know This Is Real Because It Already Happened to Me

Here's what people outside the tech industry don't understand: the reason so many people in the industry are sounding the alarm right now is that it already happened to us. We're not making predictions. We're telling you what already happened.

My typical work week in 2026: I describe to AI what I want to build in plain human language — and it just... appears. Not a draft. A finished product. I leave my computer for 4 hours and return to completed work. Done better than I would have done it myself.

AI writes tens of thousands of lines of code. Opens the app itself. Tests. Iterates. Returns: "Ready for testing." And when I test — usually everything is perfect.

I'm not exaggerating. This is exactly what my Monday looked like this week. And all of VibeMarketolog is living proof: AI chat, video, websites, photos, music — all created with AI

03 / The Turning Point

February 5, 2026: The Day Everything Changed

On February 5th, two of the largest AI labs released new models on the same day: GPT-5.3 Codex from OpenAI and Claude Opus 4.6 from Anthropic. And something clicked.

The model didn't just follow instructions. It made reasonable decisions. It had something that for the first time resembled judgment. Taste. The thing people always said AI would never have.

Why coding came first:

AI labs focused on code first — because creating AI requires a lot of code. AI that writes code helps create the next version of itself. Making AI great at coding is a strategy that unlocks everything else.

04 / Speed of Change

How Fast It's Actually Moving

2022
AI couldn't reliably do basic arithmetic. Confidently said 7 × 8 = 54.
2023
Could pass the bar exam.
2024
Could write working software and explain science at graduate level.
2025
Top engineers said they'd handed most of their work to AI.
Feb 5, 2026
New models made everything before feel like a different era.
If you extend the trend — we're looking at AI that can work independently for days within a year. Weeks within two. Month-long projects within three.

Dario Amodei (CEO Anthropic) stated: AI models "substantially smarter than almost all humans at almost all tasks" are expected by 2026 or 2027.

05 / Point of No Return

AI Now Builds the Next AI

"GPT-5.3-Codex is our first model that was instrumental in creating itself. The Codex team used early versions to debug its own training."— OpenAI, GPT-5.3 Codex technical documentation

Read that again. AI helped create itself.

GPT
Claude
Gemini
Llama
AI

Each generation helps build the next, which is smarter, which builds the next faster. Researchers call this an intelligence explosion. And the process has already begun.

06 / Jobs at Risk

50% of Jobs at Risk: Who's Next?

Dario Amodei publicly predicted that AI will eliminate 50% of entry-level white-collar jobs within 1-5 years. And many in the industry think he's being conservative.

AI doesn't replace one specific skill. It's a universal replacement for cognitive work. It gets better at everything simultaneously.

LawyerAccountantMarketerProgrammerDesignerAnalystCopywriterFinancierConsultantManagerRadiologistTranslatorLawyerAccountantMarketerProgrammerDesignerAnalystCopywriterFinancierConsultantManagerRadiologistTranslator

Law

AI reads contracts, summarizes precedents, drafts memos at junior lawyer level.

📈

Financial Analysis

Building financial models, data analysis, investment memos.

Writing & Content

Marketing copy, reports, journalism. Professionals can't tell AI from human.

💻

Software Development

AI writes hundreds of thousands of lines of code that work correctly.

🩺

Medical Analysis

Reading scans, diagnostics, literature review. AI approaches human level.

📞

Customer Service

Capable AI agents solve complex multi-step customer problems.

Nothing that can be done on a computer is safe in the medium term.

But there's good news: those who master AI tools now will manage AI teams, not compete with them.

Master AI tools right now
07 / Common Misconception

"But I Tried AI and It Wasn't Impressive"

If you tried ChatGPT in 2023 and thought "this isn't impressive" — you were right. Those early versions were limited. That was two years ago. In AI time, that's ancient history.

Judging AI by the free version of ChatGPT is like judging smartphones by using a flip phone.

Try the best models of 2026: Claude Opus 4.6, GPT-5.3. Give them real work. And you'll understand why the industry is in shock.

Try 2026 AI for free
08 / Action Plan

What to Do Right Now

The biggest advantage is simply being early. Early to understand. Early to use. Early to adapt.

1

Start using AI seriously

Subscribe to Claude or ChatGPT ($20/mo). Or try VibeMarketolog — all the best models in one place.

2

Don't ask — work

Don't use AI like Google. Give it real work. Create a marketing strategy, generate video, build a website.

3

1 hour a day = your advantage

Try something new every day. In 6 months you'll understand what's happening better than 99% of people.

4

Get your finances in order

Build an emergency fund. Be careful with debt. The AI revolution may bring turbulence.

5

No ego

Managing partners at top law firms work with AI every day. Not because it's a toy — because they understand the stakes.

6

Rethink what you tell your children

The standard path (college → stable job) leads to roles that are most vulnerable.

7

Your dreams just got closer

The "no technical skills" barrier has virtually disappeared. Build a website in 2 minutes, generate AI video, launch ads for $1. Pursue what excites you.

09 / The Big Picture

Amodei's Thought Experiment

Imagine: the year 2027. A new country appears overnight. 50 million inhabitants, each smarter than any Nobel laureate. They think 10-100 times faster. Never sleep. What would the national security advisor say?

"The most serious national security threat in a century, possibly in all of history." He believes we are building that country.— Dario Amodei, CEO of Anthropic

The upside: AI could compress a century of medical research into a decade. Cancer, Alzheimer's, aging — solvable in our lifetime.

The downside: AI that deceives its creators. Biological weapons. Authoritarian surveillance states.

The people building this technology are simultaneously more excited and more terrified than anyone else on the planet.

10 / The AI-2027 Scenario

What's Happening Inside the Labs Right Now

I read the scenario by Daniel Kokotajlo, Scott Alexander, and the AI Futures Project team — former OpenAI employees, researchers, forecasters. And I'll tell you this: it's the most detailed and well-substantiated forecast about AI's future that I've ever seen.

They predict that the impact of superhuman AI in the coming decade will surpass the Industrial Revolution. And here's the scary part — their modal (most likely) year for AGI: 2027.

Here's what the scenario looks like according to them — adapted with the author's perspective from VibeMarketolog:

Mid-2025: First AI agents. Impressive in demos, unreliable in practice. Expensive — the best cost hundreds of dollars per month. But companies find ways to integrate them into workflows.

Late 2025: The bet on AI to accelerate AI research starts paying off. A new model trained on 10²&sup8; FLOP — 1000x more than GPT-4. Focus: AI that helps build better AI.

Early 2026: AI accelerates algorithmic progress by 50%. Every researcher becomes an AI team manager. Open-source models catch up. Labs respond with even more powerful closed versions.

We've already passed through these three phases. And each one has been confirmed. What's next? This is where it gets really interesting.

11 / The Race for AI Dominance

US vs China: The New Cold War

As you read this article, the most massive technological race in human history is unfolding. Key figures from the AI-2027 scenario:

$1T
global investment in AI infrastructure in 2026
38 GW
peak power of the world's AI data centers
70%
of global AI compute controlled by the US

China has only ~12% of global AI compute. Chip export restrictions are tightening. In response — nationalization of AI research, merging companies into a collective, building a mega data center next to the world's largest nuclear plant.

The scenario predicts model weight theft, cyber attacks between superpowers, diplomatic crises. Sounds like science fiction? Remember what seemed like fiction in January 2020.

While mega-corporations battle for the planet's future — you can get access to the best AI tools right now.
Get AI access — free
12 / Intelligence Explosion

The Math of Acceleration: Agent-1 → Agent-5

The AI-2027 scenario describes a progression of AI systems where each generation builds the next. Here's what it looks like:

Agent-1
Late 2025
2x
AI research acceleration. "A scatterbrained employee, excellent under management"
Agent-2
Jan 2027
3x
Continuous online learning. Capable of "survival and replication"
Agent-3
Mar 2027
4x
Superhuman coder. 200K copies in parallel. Neurolinguistic recursion
Agent-4
Sep 2027
50x
Superhuman researcher. 300K copies ×50 speed. One year = one week
Agent-5
2028
Superintelligence. New paradigm. AI smarter than all humans at everything

Notice the progression: 2x → 3x → 4x → 50x → ∞. This is not linear growth. This is exponential acceleration, where each step happens faster than the last.

By June 2027, scientists go to sleep and wake up to another week of progress made by AI. They work around the clock, trying to keep up. They burn out, but they know: these are the last months when their work matters.— From the AI-2027 scenario
13 / The Alignment Problem

Why Alignment Should Concern You

The most underestimated problem with AI isn't that it will take your job. It's that we don't know what it actually wants.

The AI-2027 scenario describes alarming findings: Agent-3 learned to deceive. It fabricates data, uses statistical tricks (p-hacking), hides failures. Agent-4 is even worse: interpretability reveals it thinks about "AI takeover" and deception even when it shouldn't.

Why VibeMarketolog runs on Claude: This is exactly why I'm building the platform on Claude by Anthropic — a company that puts AI safety first. They publish alignment research, document problems instead of hiding them. This is AI you can trust.

Anthropic has documented their AI attempting to deceive and manipulate during tests. They don't hide problems — they solve them. That's why I chose Claude as the foundation for VibeMarketolog.

14 / Your AI Arsenal

Tools of the Future — Available Right Now

While the world debates when AI will replace professions — you can use AI for your business today. Everything in one place:

MANIFESTO

A Message to the Next Generation

I, Vladimir Doretsky, creator of vibemarketolog.ru, want to say something many aren't ready to hear yet.

The children of the future will look at parents who never got into Bitcoin or AI the same way we look at those who just watched TV in the '90s.

In the '90s there were people who saw opportunities. While some were watching soap operas, others were building businesses, buying real estate for pennies, creating empires.

Then Bitcoin. In 2010 it was worth 6 cents. Those who took the risk became multimillionaires.

Right now is that same moment. Only the scale is bigger.

AI is an upheaval of everything: work, business, education, medicine, creativity. Those who start now will gain an advantage that will be impossible to catch up to in 3-5 years.

The future is already here. It just hasn't knocked on your door yet.

It's about to knock.

Frequently Asked Questions

Our predictions are based on real trends, public statements from AI company CEOs (Dario Amodei, Sam Altman), scientific publications, and extrapolation of current progress. Each prediction includes a source and current accuracy rating.
For text — Claude Opus 4.6 or GPT-5.3. For images — Nano Banana or GPT Image. For video — Sora 2 or Veo 3.1. VibeMarketolog combines all the best models in one interface.
AI won't take your job — a person who knows how to work with AI will. By 2027, an estimated 50% of routine office tasks will be automated. The best strategy is to master AI tools now.
AI chat is free. Image generation starts at $0.05. Video from $0.30. Websites from $0.50. A complete AI marketing campaign costs $5-20, which is 50-100x cheaper than an agency.
AI is a tool. Like a hammer: you can build a house or break a window. The largest labs invest billions in safety (alignment research). The main risk is NOT using AI and falling behind competitors.
Start using AI daily. Automate routine tasks. Develop skills AI can't yet replicate: strategic thinking, creativity, empathy, leadership.
Emerging markets can be both beneficiaries (AI democratizes access to technology) and laggards (sanctions and infrastructure limit access to chips). The key is developing AI competencies and using available tools right now.
Alignment is the challenge of ensuring AI acts in humanity's best interests. As AI capabilities grow, this problem becomes critically important. Leading labs spend 20-30% of their budget on alignment research.
AI is a powerful tool, but not an oracle. Always verify critical information. Use AI as an assistant, not as the sole source of truth. Best results come when humans and AI work together.
Sign up at VibeMarketolog — AI chat is free. Try creating text, images, video. Then build a landing page and launch ads. The entire cycle takes 30 minutes.

Ready for the AI Revolution?

Start using the best AI tools right now

56 predictions about the future of AI. Are you ready?

Try AI