OpenAI's "Strawberry" Crisis: How Did an AI Model Scare the World?

September 2024. Leaked document from OpenAI: "Project Strawberry" (formerly Q*). The claim is: AI can now reason on its own, solve math problems independently of humans, and most importantly: generate its own doctrine—that is, create its own data universe.

Twitter (sorry, X) went crazy. Reddit was filled with conspiracy theories. "Artificial General Intelligence (AGI) has arrived" titles flew in technology forums.

What happened next? OpenAI quietly released the o1 model. They said "An ordinary reasoning model". They said "Not revolutionary".

But the truth is different: For the first time, an AI showed the capacity to create its own knowledge cosmos without needing humans. This is exactly what Harari calls "alien intelligence".

Harari's "Alien Intelligence" Obsession (And Why He's Not Wrong)

Yuval Noah Harari persistently says "Alien intelligence" instead of "Artificial intelligence" in his latest book Nexus. Not "artificial"; a non-biological, non-human, different kind of mind.

For the first time in history: An entity emerges that does not think like us, does not feel like us, but acts as if it thinks and decides.

If you don't detonate the nuclear bomb, it won't explode. But ChatGPT? Without anyone pressing a button, it writes tweets, generates code, produces text, argues with people, persuades. So technology acts like an agent for the first time. It is not controlled; the controlled thing now also affects the controller.

Data Colonialism 2.0: We Are Not the Product, We Are the Raw Material

We used to say "If it's free, you are the product". The reality in 2024 is darker: We are not the product. We are the raw material.

Data flow in the world: Labeling workers in Kenya teach data to ChatGPT for $2 an hour. Content moderators in the Philippines clean hundreds of traumatic images a day. People in Venezuela give brain power to AI in small drops by solving CAPTCHAs.

But the real irony: While they teach AI, AI also learns them.

Situation in Turkey? Getir, Trendyol, Yemeksepeti → shopping behavior of millions of people. Finance applications → consumption, income, risk profiles. Social media → identity, emotion, political tendency. Where is most of this data? In Seattle. In Beijing. In Dublin. Turkey's digital DNA has already been transferred abroad.

"Trust Humans, Not AI" Discourse: Is It Realistic?

Harari says: "Despite all the mistakes of history, humanity is more reliable than AI." Really?

  • Humanity resume: 20th century: 100 million war deaths. 2008: Global collapse due to the greed of a few bankers. 2020: Leaders saying "It's just the flu" → 7 million COVID deaths. 2024: Tens of thousands of civilian deaths in Gaza → the world watched.
  • AI resume: Bing Chat "fell in love with and threatened" a journalist (fixed). ChatGPT gave a fabricated case example (lawyer lost his profession). Google Gemini showed Hitler as "diverse ethnic background" (apologized).

Who is ahead in the danger race? The answer is not clear.

"Institutional Belief" Dead End: Is It the Cure or the Problem Itself?

Everyone says: "We need institutions to supervise AI." So what is the report card of existing institutions?

WHO: Late in declaring COVID a "pandemic". US SEC: Did not see FTX's $8 billion fraud. FDA: Approved OxyContin saying it is not addictive. EU: Issued GDPR, result → cookie warnings, nothing changed.

Turkey? BTK: Banned Wikipedia, everyone learned VPN. RTÜK: Tried to supervise Netflix, Disney+ came. KVKK: 100 million TR ID numbers are circulating on the dark web.

Maybe not establishing new institutions, but distributed governance is needed. Like Wikipedia. Like Bitcoin. Like Mastodon.

Immigrant Crisis vs AI Crisis: Different Package, Same Emotion

Trump: "Immigrants are eating dogs!" Tech conservatives: "AI will eat your jobs!" Same story: Fear. The Other. Invasion.

But the tragicomic side: Those producing anti-immigrant content? Right-wingers using ChatGPT. AI opponents? They talk with content carried by the Twitter algorithm.

The real "invasion" has already happened: Siri has been listening to us for 10 years. Alexa records. TikTok manipulates our minds. ChatGPT completes our thoughts. And us? We are voluntary participants.

Capitalism: The Big Variable Harari Always Misses

Harari makes a strong analysis but misses a point: The problem is not AI. The problem is capitalism itself. AI is just the latest version of capitalism.

  • Feudalism: Land → power
  • Industrial capitalism: Factory → power
  • Digital capitalism: Data → power
  • AI capitalism: Ownership of intelligence → power

Who owns intelligence today? Microsoft, Google, Amazon, Meta, Musk. 5–6 giant companies. "Democratic AI" dream? Already stillborn.

Turkey's AI Reality: Digital Vassal State

Let's be realistic: Turkey's AI strategy = "Turkifying foreign models".

Turkcell's model? Packaged version of GPT. "Domestic and National ChatGPT"? Fine-tuned version of LLaMA. TÜBİTAK's studies? Largely showcase.

Data sovereignty zero: User data → Amazon cloud. State data → Microsoft cloud. Banking data → Oracle cloud.

We are not a colony; we are a digital plantation. We produce raw data, we import processed intelligence.

Supervisory Institution Paradox: "Who Will Supervise the Supervisors?"

Institution to control AI? Great. So who will control that institution?

  • State control → China model → digital totalitarianism
  • Company self-regulation → Silicon Valley model → profit > ethics
  • International structure → UN model → endless committee, zero action
  • Anarchist approach → everyone raises their own model → freedom or chaos?

Post-Human Future: Upload or Perish?

There are two camps: "Let's merge with AI or we will perish" and "Human essence must be preserved, limits must be set".

The real camp is this: We have already merged. How many days can we live without a phone? Can we find our way without a map application? Can we write a report without ChatGPT? We are not entering the process of "becoming cyborgs". We already became.

AI Anxiety vs Climate Anxiety: Wrong Apocalypse

Everyone is afraid of the AI apocalypse. Whereas: 2024 is the hottest year in history. Amazon rainforest is on the verge of collapse. Microplastics in our blood. We are already in the 6th extinction age.

Maybe the AI panic distracts from the real issue: Capitalism destroying the planet.

Reality Crisis: Is There Such a Thing as "Truth" Left?

Influencers → Fake life, real influence. Politics → Deepfake speeches, real votes. Art → AI production, real money. Writers → AI assistance, real publication.

Deepfake crisis is on the agenda in Turkey. But the irony is: Even real videos are rejected as "deepfake". Everything is deniable. This is the name of this age: Age of Plausible Deniability.

Resistance Is Possible (But Not Simple)

  • Individual resistance: Digital minimalism, data cleaning, AI literacy.
  • Collective resistance: Open source models, cooperative platforms, ethical pressure.
  • Philosophical resistance: Reconstructing the definition of humanity, creating a new ethic, imagining a post-capitalist future.

Realistic Emergency Plan for Turkey

Data Sovereignty

  • Pull critical infrastructure from foreign clouds
  • Establish domestic data centers
  • Regulate data export

Social Awareness

  • "AI and Critical Thinking" course from primary school
  • "Not every ChatGPT answer is correct" public service announcement
  • Deepfake literacy

Strategic Positioning

  • Competing with GPT → you lose
  • Find niche area: Ottoman archive, Turkish language models
  • Become a regional AI center

Defensive Democracy

  • Limitations on AI use during election period
  • Heavy penalty for Deepfake
  • Identity requirement for bot accounts

Big Twist: What If Harari Is Also Wrong?

Maybe the issue is not: "Human vs AI". Maybe the real war is: Central Power vs Distributed Creativity.

Facebook's human moderators are also problematic. China's AI surveillance too. OpenAI's company-oriented models too. Anonymous hackers' own bots too.

The problem is not technology. The problem is power. It always was.

Final: The Dance Continues

I asked this before: "Are you with the algorithms or humanity?" Now I ask a harder question: Is there really a difference left between the two?

We are dancing with AI. Sometimes it leads, sometimes we do. Sometimes we are in harmony, sometimes our feet get tangled. The dance doesn't end. Only the music changes.

GPT-3 was a waltz. GPT-4 tango. GPT-5? Maybe breakdance. Maybe a fight.

The important thing is not to lose the rhythm. The important thing is not to forget who chose the music. The important thing is to claim one thing: Agency. Human with will.

Don't forget: We can change the dance if we want. We can change the music if we want. We can invent a completely new dance if we want.

If we still have a choice... The choice is ours.