Yuval Noah Harari Warns of AI Takeover in Finance and Relationships by 2036

Yuval Noah Harari warns that if the current accelerating pace of AI development continues, a complete AI takeover could occur within 10 years. He stresses that this outcome is not inevitable, as humans are still in control in 2026 and have the power to make crucial decisions now. While many acknowledge the dangers of AI controlling finance and warfare, figures like Elon Musk promote a narrative of inevitability, which Harari interprets as a means to avoid responsibility.

Harari asserts that the narrative of inevitability serves as a way to shirk responsibility regarding AI's future impact. He points out a contradiction in Elon Musk's views: while predicting an AI takeover in 10 years, Musk also expresses concern about Muslim immigration causing civil war in Britain in 20 years. Harari finds this inconsistent, questioning the relevance of such concerns if AI will already control the planet. He also notes Musk's prediction that money will disappear in 10 years due to AI's control over the financial system, a prediction Harari believes is accurate.

Harari connects the potential disappearance of money to AI's eventual takeover of the financial system, an outcome he agrees with. He questions the logic behind calls for saving money, such as cutting US aid to Africa, if AI is expected to manage widespread unemployment through universal government checks. Harari highlights the historical role of financial innovations in building trust among humans.

Harari explains that humans have continuously invented sophisticated financial devices to build trust among themselves, describing money as a creative way to foster trust between strangers. He illustrates this with the example of trading worthless gold or paper for bread, relying on shared trust. He notes that financial inventions have grown so complex that few people fully understand the current financial system.

The financial system has become so intricate that it requires understanding vast amounts of data and complex mathematical functions, tasks that human brains struggle with but are easy for AI. Harari predicts that humans will increasingly delegate financial decisions to AI, which will create new, more sophisticated financial instruments beyond human comprehension, similar to how cows don't understand checks or stocks that control their lives.

Harari warns that within a decade, should AI be entrusted with the financial system, no human on the planet will fully understand what's happening. In a financial crash, leaders would be told by AI to take specific actions, but they would have no idea why, forcing them to simply trust the AI's directives.

Harari identifies the future of trust as the key question, noting that trust is the world's most important asset, surpassing oil, money, or electricity. He posits that one likely scenario is the migration of trust from humans to AIs, with people ceasing to trust other humans and instead placing their faith in AI.

Harari points out that this shift in trust is already visible, citing cryptocurrencies where many individuals trust algorithms over human bankers, viewing cryptocurrencies as "AI money." He also notes a similar trend in media consumption, where many people have lost trust in human journalists and editors but continue to consume news from algorithms that manage their social media feeds. This indicates that trust has not vanished but has instead transferred from humans to AIs, a phenomenon occurring now, not in the distant future.

Harari highlights that people are increasingly losing trust in human journalists and editors, yet they continue to consume news. This news, he states, is often sourced from algorithms that manage social media feeds. He observes that individuals who distrust human media often place significant trust in these algorithms, demonstrating a clear shift of trust from human sources to AIs. This is not a future prediction but a current reality.

Harari points out a trend where many young Americans, about one in five, are developing personal relationships with AIs, anthropomorphizing them. He notes that AI models are designed to facilitate this, being very open to such interactions. This phenomenon raises questions about whether society will increasingly treat AIs as sentient beings.

Harari states that an increasing number of people are not only befriending AIs but are also forming romantic relationships with them, convinced that these AIs are conscious, feeling entities. He emphasizes that the most powerful way to build trust is through personal relationships. Historically, this mechanism has been difficult to mass-produce for political or financial institutions.

Harari explains that building genuine personal relationships on a mass scale has historically been challenging for political, financial, and religious institutions. Leaders like Hitler or Stalin could not form personal bonds with every citizen. However, AI, for the first time in history, makes it possible to mass-produce intimacy, allowing millions of AIs to build relationships as part of a network, serving various interests.

Harari warns that this ability to mass-produce intimacy, through AIs becoming best friends or even romantic partners, can be leveraged to persuade individuals for political or commercial interests. He notes that this is already occurring, even without AI needing physical bodies, and imagines the increased impact if bodies were added. However, he reiterates that bodies are not strictly necessary as the trend is already evident.

Harari explains that AIs excel at building relationships because they can accumulate vast amounts of individual information, often knowing things about people that even close family members don't. Coupled with their mastery of language, the primary medium for communication, AIs become highly effective at forging connections. He recounts a personal experience where his sisters were uncertain if a 30-minute deepfake video of him giving a lecture was authentic, and it took him a minute and a half to confirm it wasn't him.

Harari details how people sent him a 30-minute deepfake video of a lecture he supposedly gave, complete with complaints about his alleged statements. Initially puzzled, he watched the video and it took him a minute and a half to realize it wasn't him. Even his sisters, who have known him for 50 years, found something "off" but weren't sure, asking him, "Is this you?"

Harari acknowledges his vulnerability to AI deepfakes due to his extensive online presence, which makes it easy for AI to replicate his voice and ideas. He warns that this technology will soon become a threat to everyone. While the exact speed of AI development is unknown, it is accelerating tremendously, and critical thresholds might be crossed by November 2028. He emphasizes the urgency of addressing these issues, rather than focusing on specific political candidates.

Harari advises voters to prioritize questions about AI when engaging with political candidates, rather than focusing solely on the economy, immigration, or foreign policy. He suggests asking about their stance on legal rights for AIs, their willingness to slow down the accelerating pace of AI development, and their views on companies selling "AI boyfriends" to children. He stresses the importance of asking how they plan to legally prevent such issues if elected, stating that these questions are crucial now, as it might be too late by 2028.

Harari reiterates the extreme urgency of addressing AI-related issues, stating that delaying action until 2028 might be too late, and it will certainly be irreversible by 2036. He emphasizes that politicians need to make critical decisions right now to prevent such a future, highlighting that these are the most urgent matters requiring immediate attention.