Harari: artificial intelligence can dominate humanity – 03/28/2023 – Tech

Imagine that as you board a plane, half of the engineers who built it tell you that there is a 10% chance it will crash, killing you and everyone else on board. Would you still board?

In 2022, more than 700 academics and researchers linked to leading artificial intelligence companies were interviewed about the future risk of AI. Half of respondents stated that there was a 10% or greater chance of human extinction (or equally permanent and severe loss of power) by future AI systems. The technology companies that build today’s great models of language are in a race to get all of humanity on that plane.

Pharmaceutical companies cannot sell new drugs to people without first putting their products through rigorous safety checks. Biotech labs can’t release new viruses into the public sphere to impress shareholders with their magic. Likewise, AI systems with the power of GPT-4 and beyond must not be involved in the lives of billions of people at a faster rate than cultures can safely absorb them. A race to dominate the market should not define the speed of deployment of humanity’s most important technology. We must move at the speed that allows us to do this right.

The specter of AI has haunted humanity since the mid-20th century, but until recently it remained a distant prospect, something that belongs more in science fiction than in serious scientific and political debate. It’s hard for our human minds to comprehend the new capabilities of GPT-4 and similar tools, and it’s even harder to comprehend the exponential speed at which these tools are developing even more advanced and powerful capabilities. But most of the key skills come down to one thing: the ability to manipulate and generate language, whether it’s words, sounds or images.

“In the begining was the verb.” Language is the operating system of human culture. From language emerge myths and laws, gods and money, art and science, friendships and nations – even computer codes. The AI’s new mastery of language means that it can now hack and manipulate civilization’s operating system. By gaining mastery of language, AI is seizing the master key of civilization, from bank vaults to holy tombs.

What would it mean for humans to live in a world in which a large percentage of stories, melodies, images, laws, policies and tools are shaped by a non-human intelligence, which knows how to exploit with superhuman efficiency the weaknesses, prejudices and vices of the mind – while also knowing how to form intimate relationships with human beings? In games like chess, no human can expect to beat a computer. What happens when the same thing happens in art, politics and even religion?

AI could quickly consume all of human culture – everything we’ve produced over thousands of years – digest it and start producing a flood of new cultural artifacts. Not just school essays, but also political speeches, ideological manifestos, and even holy books for new cults. In 2028, the US presidential race may no longer be run by humans.

Humans often don’t have direct access to reality. We are encapsulated by culture, experiencing reality through a cultural prism. Our political views are shaped by reports from journalists and anecdotes from friends. Our sexual preferences are shaped by art and religion. That cultural cocoon has so far been woven by other humans. What will it be like to experience reality through a prism produced by a non-human intelligence?

For thousands of years, we humans have lived inside other humans’ dreams. We worship gods, pursue ideals of beauty, and dedicate our lives to causes that originated in the imagination of some prophet, poet, or politician. Soon, we too will find ourselves living inside the hallucinations of non-human intelligence.

The “Terminator” franchise featured robots running through the streets and shooting people. “The Matrix” assumed that in order to have full control of human society, AI would first have to gain physical control of our brains and connect them directly to a computer network. In reality, however, simply by mastering language, AI would have everything it needs to contain us in a Matrix-like world of illusions, without shooting anyone or implanting chips in our brains. If a shot is needed, AI can make humans pull the trigger, just by telling us the right story.

The specter of being trapped in a world of illusions has haunted humanity for much longer than the specter of AI. Soon we will finally be face to face with Descartes’ demon, with Plato’s cave, with the Buddhist Maya. A curtain of illusions could descend over all of humanity, and we may never be able to pull it back – or even realize it’s there.

The social network was the “first contact” between AI and humanity, and humanity lost. The “first contact” gave us the bitter taste of what is to come. In social networks, primitive AI was used not to create content, but to curate user-generated content. The AI ​​behind our news feeds keeps choosing which words, sounds and images hit our retinas and eardrums, based on selecting the ones that will get the most virality, the most reaction and the most engagement.

Although very primitive, the AI ​​behind the social network was enough to create a curtain of illusions that increased social polarization, undermined our mental health and destroyed democracy. Millions of people have confused these illusions with reality. The United States has the best information technology in history, but American citizens can no longer agree on who won elections. Although everyone is already aware of the negative side of the social network, it has not been addressed, because many of our social, economic and political institutions have become involved with it.

Large language models are our “second contact” with AI. We can’t afford to lose again. But on what basis are we to believe that humanity is capable of using these new forms of AI to our advantage? If we continue as usual, new AI capabilities will once again be used for profit and power, even as they inadvertently destroy the foundations of our society.

AI really does have the potential to help us defeat cancer, discover life-saving drugs, and invent solutions to our climate and energy crises. There are countless other benefits that we cannot even imagine. But never mind how high a skyscraper of benefits AI can accumulate if the foundation crumbles.

The time to rely on AI is before our politics, our economy and our daily lives become dependent on it. Democracy is a conversation, the conversation depends on language, and when language itself is hacked, the conversation stops and democracy becomes unsustainable. If we wait for chaos to set in, it will be too late to remedy it.

But there is a question that may hover in our mind: “If we don’t move as quickly as possible, won’t the West risk losing to China?” No. The implantation and uncontrolled integration of AI in society, freeing semi-divine powers dissociated from responsibility, could in fact be the reason why the West loses to China.

We can still choose which future we want with AI. When semi-divine powers are combined with proportionate responsibility and control, we can realize the benefits that AI promises.

We summon an alien intelligence. We don’t know much about it, except that it’s extremely powerful, it offers us wonderful gifts, but it could also sabotage the foundations of our civilization. We call on world leaders to respond to this moment at the level of challenge it represents. The first step is to buy time to upgrade our 19th century institutions for a post-AI world and learn to master AI before it dominates us.