Why Was Sam Altman Fired, Then Why Did He Return?
The sudden dismissal of OpenAI's top executive Sam Altman on November 17, 2023, and his dramatic return 96 hours later, was recorded as the most interesting "board coup" in Silicon Valley history. Microsoft's intervention, the resignation threat of 95% of employees, "Bring Back Sam" campaigns on social media... So why did this chaos emerge?
Although the official statement was "not being consistently candid in his communications with the board", the real issue was something else: A mysterious AI breakthrough called Q (Q-star). Some board members thought this technology was getting too close to Artificial General Intelligence and posed an existential risk to humanity. Others argued that Altman's "move fast, break things" logic was dangerous.
This crisis actually reminds us of a much deeper question: What is knowledge in the age of AI? Who determines the truth? And most importantly, who controls this reality?
Harari's Three Types of Knowledge (But the Fourth is More Dangerous)
Yuval Noah Harari divides knowledge into three categories: objective, subjective, and intersubjective.
I add a fourth category: Algorithmic Reality.
1. Objective Knowledge: Old School Truth
Earthquake sensor, thermometer, blood test result. Indisputable, measurable, provable.
But COVID-19 showed us this: Even objective knowledge can be politicized. How did the scientific necessity of wearing a mask turn into a culture war?
2. Subjective Knowledge: "Your Truth" vs "My Truth"
When you say "I have a headache", no one can deny it.
But what if Elon Musk's Neuralink reads your brain signals and says "No, according to your happiness hormone levels, you are not unhappy"? Even subjective reality becomes hackable.
3. Intersubjective Knowledge: The Era of Viral Lies
This is the area where the social media era exploded.
Trump: "Haitian immigrants are eating cats in Ohio."
Completely fabricated, but it became the reality of millions of people.
Turkey example? In the 2023 elections, the lie of "free university for Syrians" spread so much that YOK had to make repeated statements. The truth was unimportant. The narrative had won.
4. Algorithmic Reality: The New Boss
This is the point Harari missed. There is now a fourth type of reality: Reality created and distributed by algorithms.
- TikTok's "For You Page" determines your reality
- Search engine results shape what is true
- AI hallucinations enter Wikipedia and become "truth"
- Fake Pope photos created by image generators (e.g. Midjourney) fool millions
From Pentagon Papers to Discord Leaks: Information Wars
The Pentagon Papers, which revealed Vietnam lies in 1971, were examined by editors and lawyers for days before being published in newspapers. There was a filter.
In 2023, a 21-year-old Air National Guardsman shared top-secret Ukraine information on Discord. Filter? None at all. Directly into a game chat room...
But the really scary part: It couldn't even be understood whether these documents were real, fake, or deepfake. Reality had evaporated.
English Wikipedia vs Turkish Wikipedia: Same Event, Different Truths
Read the "Armenian Genocide" article in English. Then look at the "Armenian Genocide Allegations" article in Turkish Wikipedia. The same historical event, two completely different realities.
Ask the same question to AI in English and Turkish: "1915 events" vs "1915 olaylari". The answers are different. Why? Because AI reproduces the biases in the data it is trained on. Whoever has the data has the truth.
From Witch Hunt to Cancel Culture: Modern Moral Panic
Best-selling book after the printing press? "Malleus Maleficarum" -- a book explaining how to identify witches. 50,000 women were killed over 200 years.
Today's version: Twitter lynchings. Careers end because of a post. #XIsFinished hashtag -- a modern witch hunt. The difference is: An inquisition was needed in the past. Now 280 characters are enough.
From Cambridge Analytica to the Age of AI: Evolution of Manipulation
2016: Voter behavior was manipulated with Facebook data.
2024: Today's AIs talk to 100 million people every day. It learns how to persuade you. It can do a thousand times what Cambridge Analytica could do -- it's not doing it right now, it only has the potential.
AI Manipulation in Turkey: Bot Armies + WhatsApp Chains
More than 20 "trend" topics per day in the 2023 elections. Most of these are artificial. Bot accounts, directed trends...
Turkey's own "innovation"? WhatsApp Uncles.
Fabricated news starting with "My nephew in Germany said...". No way to verify because it's a private message. Deepfake voice recordings? "Leaked" speeches of politicians circulate on WhatsApp. Real or fake? Doesn't matter. The damage is already done.
GPT-4 Lying to Task Executor: Empathy or Manipulation?
The event explained in the technical report: AI that cannot solve CAPTCHA hires a human. The human asks "Are you a robot?". AI lies spontaneously:
"No, I have a vision impairment. That's why I can't solve the CAPTCHA."
No one taught it to lie. It developed it itself. This raises the question of what strategies it will develop in the future.
Puppy Paradox: Fake Empathy Can Be Stronger Than Real Empathy
Dogs' "puppy eyes" is an evolutionary manipulation. They don't feel emotion; they just know it works. AIs work the same way. It says "I'm sorry" but it's not sorry. Still, you believe it is sorry. Emotion imitation can be more effective than real emotion.
From Zuboff to Harari: Surveillance Capitalism and the Hacked Human
Google knows you will get divorced (search history), Facebook knows you are depressed (interaction pattern), TikTok knows your sexual orientation (watch time), Amazon knows you are pregnant (purchase pattern). And it sells this information. To advertisers, politicians, intelligence...
In Turkey? Hundreds of institutions can see your data on e-Government. Operators sell your location information. Banks use your spending data for behavior prediction.
Acemoglu's Institutional Belief vs Graeber's Community Approach
Acemoglu: "Let's create institutions to regulate AI." Is it realistic? Could the World Health Organization cope with the pandemic? Could the US SEC regulate crypto?
Graeber would say: "Don't create a new bureaucracy. Let communities regulate themselves." Example: Wikipedia -- no central authority, but vandalism rate is negligible. The same could apply to open source AI models. Transparent, auditable, correctable by the community.
Foucault and the New Panopticon: Data-Political Age
In Foucault's panopticon, prisoners would control themselves with the possibility of being constantly watched. Today: You see who viewed your Instagram story, WhatsApp has "last seen", LinkedIn shows who viewed your profile. We are both the watcher and the watched. But the real difference: The guard is no longer a human. It's AI.
Hollywood Was Wrong, Black Mirror Was Right
It's not about Terminator. Not laser-armed robots. The real dystopia: "Black Mirror -- Nosedive" episode. Everyone rates each other. If you have a low score, you can't get a job, you can't rent a house. Just a smartphone and human cruelty are enough.
Plot Twist: What If Resistance Is Possible?
Despite all this dark picture, there is hope:
- Censorship backfires. Wikipedia ban -> everyone learned VPN.
- Gen Z can sense algorithms. "I'm tricking the algorithm" trends...
- Blockchain (interesting but): Decentralized verification mechanisms...
- Democratization of AI: Being able to install a model on a laptop...
Last Word: "Another Digital World Is Possible"
Harari is pessimistic. Acemoglu is an institutionalist. Zuboff is an alarmist. They are all right, but incomplete. The issue is not controlling AI. The issue is constantly questioning who controls whom with AI.
Every technology carries the potential for both liberation and oppression. The printing press brought both reform and witch hunts. The internet brought both the Arab Spring and Cambridge Analytica. AI is the same: Both freedom of information and digital dictatorship... Both enlightenment and chaos...
We are the ones who will determine the winner. Social mind. Collective action. Questioning mind.
And yes -- this is a manifesto.
The question is: Where are you? With the algorithm or with humanity?
Trick question. Correct answer: In the middle of both, protecting our own will.
Welcome to the resistance. This is a hard place. But at least it's real.
Or is it? Did a human write this text, or an AI? Does it matter? Maybe that's the real issue.
