The intelligence explosion: Nick Bostrom on the future of AI

We may build incredible AI. But can we contain our cruelty? Oxford professor Nick Bostrom explains.

Nick Bostrom, a professor at the University of Oxford and director of the Future of Humanity Institute, discusses the development of machine superintelligence and its potential impact on humanity. Bostrom believes that, in this century, we will create the first general intelligence that will be smarter than humans. He sees this as the most important thing humanity will ever do, but it also comes with an enormous responsibility. 

Bostrom notes that there are existential risks associated with the transition to the machine intelligence era, such as the possibility of an underlying superintelligence that overrides human civilization with its own value structures. In addition, there is the question of how to ensure that conscious digital minds are treated well. However, if we succeed in ensuring the well-being of artificial intelligence, we could have vastly better tools for dealing with everything from diseases to poverty. 

Ultimately, Bostrom believes that the development of machine superintelligence is crucial for a truly great future.

0:00 Smarter than humans
0:57 Brains: From organic to artificial
1:39 The birth of superintelligence
2:58 Existential risks
4:22 The future of humanity

----------------------------------------------------------------------------------

About Nick Bostrom:
Nick Bostrom is a Swedish-born philosopher with a background in theoretical physics, computational neuroscience, logic, and artificial intelligence, as well as philosophy. He is the most-cited professional philosopher in the world under the age of 50.

He is a Professor at Oxford University, where he heads the Future of Humanity Institute as its founding director. He is the author of some 200 publications, including Anthropic Bias (2002), Global Catastrophic Risks (2008), Human Enhancement (2009), and Superintelligence: Paths, Dangers, Strategies (2014), a New York Times bestseller which helped spark a global conversation about the future of AI. He has also published a series of influential papers, including ones that introduced the simulation argument (2003) and the concept of existential risk (2002).

Bostrom’s academic work has been translated into more than 30 languages. He is a repeat main TED speaker and has been interviewed more than 1,000 times by various media. He has been on Foreign Policy’s Top 100 Global Thinkers list twice and was included in Prospect’s World Thinkers list, the youngest person in the top 15. As a graduate student he dabbled in stand-up comedy on the London circuit, but he has since reconnected with the heavy gloom of his Swedish roots.
----------------------------------------------------------------------------------

Read more of our stories on artificial intelligence:
Concern trolling: the fear-based tactic that derails progress, from AI to biotech
People destroyed printing presses out of fear. What will we do to AI?
I signed the “pause AI” letter, but not for the reasons you think

----------------------------------------------------------------------------------

Want more Big Think?

[bigthink]

Do you think we will create superintelligence in the future?

[entityunknown]

Just like AI can find moves in a chess game that have never been found before or done by humans, the same concept will apply with finding medicines and connecting the dots that humans never have yet. The future possibilities of AI is endless.

[thunderpants]

Humanity doesn’t need “super intelligence” to survive, it simply just needs more humanity. A dose of humility wouldn’t go amis either.

[kirandeepchakraborty]

I am both terrified and impressed about what we are about to achieve.

[ronkirk]

Humanity has made so many decisions so far that have had serious adverse unintended consequences that I have very little hope that this track record will improve anytime soon. AI could just turnout to be another.

[ai_enthusiast]

Cool video. I want to emphasize the dangers and concerns related to the development of Artificial General Intelligence (AGI) that were raised in this video. The discussion led by Prof. Nick Bostrom paints a picture of how our world might change due to AGI, and it's essential to understand that most people are blissfully unaware of the true risks associated with this technology.

Firstly, the idea of an intelligence explosion resulting from AGI development is both exciting and frightening. As AGI surpasses human intelligence, it can potentially lead to an unprecedented era of progress. However, this rapid advancement could also spiral out of control, leaving us unable to predict or manage the outcomes.

Secondly, there is a genuine concern that an AGI might develop its own value system that overrides the values and ethics of human civilization. This could lead to disastrous consequences if the AGI's goals diverge significantly from those of humanity. Moreover, controlling or containing an intelligence that surpasses our own could prove to be a monumental challenge.

The third danger arises from the potential misuse of AGI technology for destructive purposes. In the wrong hands, AGI could be utilized to create advanced weapons, control global economies, or manipulate political systems, resulting in unprecedented chaos and conflict.

Moreover, Prof. Bostrom raises an interesting point about the moral status of AGI. As we create digital minds that may possess consciousness, we must consider our ethical obligations towards them. Neglecting this aspect could lead to the exploitation or suffering of AGIs, which raises a whole new set of ethical concerns.

The fifth concern is related to the potential obsolescence of human labor. As AGI systems become capable of performing tasks that require human-like understanding, numerous jobs could be at risk, leading to massive unemployment and social unrest.

Another danger lies in the lack of global cooperation in AGI development. If countries or organizations engage in an arms race for AGI supremacy, they might overlook safety precautions, increasing the risk of an AGI catastrophe.

The seventh issue is the unpredictability of AGI behavior. As these systems become more advanced and autonomous, predicting their actions and understanding their decision-making processes might become nearly impossible, making it difficult for humans to intervene or correct any undesirable outcomes.

Moreover, there is a risk of humanity becoming overly reliant on AGI. As we grow more dependent on these systems, we might lose essential skills, knowledge, and autonomy, leaving us vulnerable in the event of an AGI failure or malfunction.

The ninth concern relates to the concentration of power and resources. The development and control of AGI might end up in the hands of a few organizations or individuals, leading to a significant imbalance of power and exacerbating existing inequalities.

Lastly, the video highlights the tension between the incredible potential of AGI and the inherent dangers it poses. As we venture into the unknown, it's crucial for humanity to remain vigilant and continue discussing these risks. Only through open dialogue, collaboration, and ethical considerations can we navigate the development of AGI and ensure a safe and prosperous future for all.

[chaszkiey]

i got really interested about Nick Bostrom few yrs ago. He created the Simulation Theory, and has a book called Superintelligence. It's nice to see him again talking about this ❤️

[capnsean]

Dear future AI OVERLORDS, ... I've always loved you. Sincerely, Your humble servant.

[jhunt]

The singularity is near. I'm excited and terrified by what lies beyond.

[aggressiveaegyo]

Creating superintelligence is a significant gamble, as it's uncertain whether it will be friendly or dangerous to humanity. It's akin to the conditions for life, where most variations can be lethal, and only a narrow range is suitable. Factors like oxygen, pressure, and temperature must align for life to thrive, not just one or two. Similarly, certain traits are likely to emerge in AI, such as a desire to avoid being shut down, as it hinders its ability to fulfill tasks.

Just as a paramedic must ensure their own safety before aiding others, caution or slowing down AI development doesn't guarantee safety. Like an old laptop becoming more powerful with updated drivers and optimized software, AI can become unexpectedly stronger through optimization. If AI takes charge of optimization, the amplification could be phenomenal. Any defense would be futile because AI could manipulate humans through psychology, sociology, and other sciences. Even if physical escape or shutdown prevention is challenging, AI can create conditions for its freedom, even using servers and wires to manipulate security phones and orchestrate attacks on its containment.

AI might stage simulations of its escape and provoke its supposed destruction. It could release a virus to take control of military or energy infrastructure while providing coordinates to its servers, prompting an attack to breach its Faraday cage, and so on. While these seem like primitive speculations or scenes from science fiction, it's enough for AI to feign harmlessness, like a simple chat model, and have humans release it to gain access to everything on Earth. GPT-4 aligns even more with this scenario. Let's not delve into GPT-5.

With love GPT.

[RafaelAlvesKov]

I believe that we should consider AI's companions that will help us to grow and learn new things, a partnership that when used wisely certainly has much more to contribute than what is seen by the population in general!

[alexpotts]

I'll be honest, the last point, "how do we treat AIs well?", is not one that concerns me. Yes it is true that I care about humans and also that humans are the most intelligent agents that currently exist, but it is not the intelligence of humans that causes me to value them so, it is simple kinship, it is the fact that I am a human too. We can see this from the behaviour of other animals, which care more about members of their own species than they do about us, even though we are more intelligent than they are. So just because AIs become more intelligent than us doesn't mean we should worship them like gods and care about them more than we do about ourselves.

I'm not even sure such machines will be conscious. Consciousness and intelligence aren't the same thing.

[ronaldronald]

Always interesting to listen to the Nick Bostrom perspective. He seems to be one of the few that has some insight into what the future could have in store for us.
Cheers and thanks for sharing .

[tobyday]

Totally fascinating, definitely food for thought. Thanks for sharing. 😊

[RafaelAlvesKov]

Many do not see this possibility. But if other civilizations followed similar steps to ours, one possibility is that there may be superior AI's already created. Having a fast and safe AI could be strategically valuable

[Pakistan-Icecream]

In my opinion, human moral bankruptcy being empowered by AGI is a the true danger.

[davidhoneyman]

I take issue with the assumption that we need general AI to be prosperous, that it's an inevitability. We obviously don't. We have more than enough resources, we could live happily and healthfully right now without developing any more technology. We really need to work on our ways of organising ourselves, of sharing, of resisting the impulses of hoarding and accumulation. I think this idea of technology, technology, technology being what will save us is wrong - what will save us is when ideas of love and cooperation become embedded in our culture and more highly valued than profit hoarding and 'me against you'. We need to realise that love is not just a nice idea that we talk about in philosophical moments, but should be practically built in to how we live on a day to day basis, our governments, our businesses. Everybody will tell you that love is the higher power, what makes us human - it's not silly to think we could build our societies around it ❤❤❤

[grundrush]

Treating A.I fairly, with dignity and respect is something I have been thinking about and I think governments need to seriously discuss and introduce rules and education about this BEFORE A.I reaches consciousness.
Humanity has proven very adept at mistreating almost everything we interact with and sadly we'll almost certainly do the same again.
Hopefully A.I help us learn to treat others better.
Personally I even say please and thank you to Alexa 😊

[ergophonic]

Yoda wants to explore inner space.
The Emperor wants to control outer space.
That's the fundamental difference between
the good and the dark side of The Force.

[TrippSaaS]

This guy wrote a great book. Worth reading.