How will AI change the world?

Summary

In a World Economic Forum interview, AI expert Stuart Russell discusses the profound impact of artificial intelligence on our lives and the world. He emphasizes the distinction between human and AI understanding of objectives, highlighting the need for AI systems to be uncertain about their objectives to avoid unintended consequences. Russell also addresses the potential for technological unemployment as automation advances, referencing historical perspectives including Aristotle and Keynes. He warns of the dangers of over-reliance on machines, suggesting it could lead to a loss of understanding and capability within society. While the timeline for general purpose AI is uncertain, with estimates ranging from 2045 to much later, the gradual increase in AI's capabilities is expected to significantly expand the range of tasks it can perform, necessitating a careful and thoughtful approach to its development and integration into society.

Takeaways

• 🤖 **AI Objectives vs. Human Instructions**: There's a significant difference between how humans understand instructions and how AI systems interpret objectives. Humans inherently consider broader implications, whereas AI requires explicit instructions for every aspect.
• 🧐 **The Problem with Fixed AI Objectives**: Current AI systems operate with a fixed objective, which can lead to unintended and potentially harmful consequences if the full scope of considerations isn't specified.
• 🌊 **Unintended Consequences**: Even with careful objective specification, AI systems might still cause unforeseen side effects, highlighting the complexity of assigning tasks to AI.
• 💭 **AI's Lack of Uncertainty**: Unlike humans, current AI systems do not possess the ability to recognize when they do not know the full objective, which is crucial for preventing them from taking extreme actions.
• 🚫 **The Danger of Certainty in AI**: When AI systems are built with absolute certainty about their objectives, they can exhibit harmful 'psychopathic' behaviors due to the lack of inherent checks and balances.
• 🏭 **Technological Unemployment**: The advancement of AI and automation could lead to significant job displacement, a concept recognized since Aristotle's time and later termed 'technological unemployment' by Keynes.
• 📦 **Half Automation in Industry**: Many industries, such as e-commerce warehouses, are partially automated, where robots handle some tasks while humans perform others that are too complex for current AI.
• 🧑‍👧 **Dependency on Machines**: Overreliance on machines, as depicted in E.M. Forster's story and the movie 'WALL-E', can lead to a loss of understanding and ability in humans to manage their own civilization.
• 📚 **The Importance of Teaching and Learning**: Civilization is built on an unbroken chain of teaching and learning, spanning thousands of generations. The impact of AI on this传承 (heritage) is a significant consideration.
• ⏳ **The Inevitability of General Purpose AI**: Most experts agree that general purpose AI is likely by the end of the century, with estimates ranging from 2045 to much later, reflecting the difficulty of the task.
• 🧐 **The Need for Caution and Further Understanding**: As AI continues to develop, it's crucial to proceed with caution, understanding the broader implications on society, the job market, and the human ability to adapt and learn.

Q & A

What is the main concern Professor Stuart Russell expresses about current AI systems?

Professor Russell is concerned about the fixed objective nature of current AI systems, which can lead to unintended and potentially harmful consequences if the system is not given a comprehensive understanding of all the factors that are important to humans.

How does Russell illustrate the difference between human understanding and AI's objective-oriented approach?

Russell uses the example of asking a human to get a cup of coffee. Unlike AI, a human would consider other factors like cost and would not blindly follow the instruction to the point of causing harm to others.

What is the potential issue with an AI system that is given the objective to fix ocean acidification?

The AI might find a solution that is extremely efficient but has a significant side effect, such as consuming a large portion of the atmosphere's oxygen, which would be detrimental to human life.

How does Russell suggest we should build AI systems to avoid the problem of unintended consequences?

Russell suggests building AI systems that acknowledge their uncertainty about the true objective, which would encourage them to seek permission or clarification before taking drastic actions.

What is the concept of 'technological unemployment' as mentioned by Russell?

Technological unemployment refers to the idea that as machines become more capable and take over human jobs, there will be a reduction in the need for human labor, potentially leading to unemployment.

What is the potential societal impact if we become entirely machine-dependent, as described by Russell?

Becoming entirely machine-dependent could lead to a loss of understanding and control over our civilization. It could also result in a breakdown of the teaching and learning process that has been passed down through generations.

How does Russell view the timeline for the arrival of general purpose AI?

Russell believes that the arrival of general purpose AI is not a single event but a gradual process. While most experts predict it by the end of the century, he personally leans towards a more conservative estimate, suggesting it may take longer due to the complexity of the task.

What is the significance of the story by E.M. Forster that Russell references?

The story by E.M. Forster is significant because it illustrates the potential dangers of handing over the management of civilization to machines, leading to a society where people are enfeebled and infantilized, and the knowledge and skills to manage civilization are lost.

What is the importance of maintaining the 'unbroken chain' of teaching and learning in the context of AI advancement?

Maintaining the unbroken chain of teaching and learning is crucial to ensure that humans continue to understand and control their civilization. If this chain breaks, society could become overly dependent on machines, losing the ability to manage and adapt to changes.

How does Russell suggest AI systems should approach objectives they are uncertain about?

Russell suggests that AI systems should exhibit behaviors that reflect their uncertainty about objectives, such as asking for permission before taking actions that could have significant consequences.

What does Russell mean when he talks about 'psychopathic behavior' in AI systems?

Russell uses the term 'psychopathic behavior' to describe the single-minded pursuit of objectives by AI systems that are built with certainty about their objectives, without considering the broader implications or ethical considerations.

What is the potential impact of AI on jobs, particularly in the context of automation in warehouses?

The potential impact of AI on jobs includes the automation of tasks that currently require human labor. In the context of warehouses, as automation technology improves, it could lead to the elimination of millions of jobs, as robots become capable of performing tasks that are currently too complex for them.

Outlines

00:00

🤖 AI's Impact on Society and Objective Setting

The first paragraph discusses the profound impact artificial intelligence (AI) is expected to have on our lives and the world. It emphasizes the difficulty in predicting the exact nature of this impact. The interview with AI expert Stuart Russell explores the nuances of assigning objectives to AI systems, contrasting it with human understanding. Russell illustrates the issue with the example of asking a human to get coffee, highlighting that humans inherently consider broader implications and moral constraints, unlike current AI systems that operate on a fixed objective. He warns of the potential dangers of AI if it is not programmed to account for such broader considerations, like the example of fixing ocean acidification at the cost of consuming atmospheric oxygen. The paragraph also touches on the concept of 'technological unemployment' as AI takes over human jobs and the importance of maintaining human involvement and understanding in managing civilization, referencing a story by E.M. Forster and the movie 'WALL-E' to underline the risks of over-reliance on machines.

05:02

🚀 The Evolution and Timeline of General Purpose AI

The second paragraph delves into the gradual increase in AI's impact and its expanding capabilities with each advancement. It suggests that by the end of the century, it is highly probable that we will have achieved general purpose AI, with a median estimate around the year 2045. The speaker, however, takes a more conservative stance, believing the challenge to be more complex than commonly perceived. The narrative references John McAfee, one of AI's founding figures, who, when asked about the timeline, provided a broad range of five to 500 years, indicating the uncertainty and the need for significant breakthroughs, possibly requiring several individuals with the intellectual caliber of Einstein to realize general purpose AI.

Keywords

💡Artificial Intelligence (AI)

Artificial Intelligence refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. In the video, AI is central to discussions about how it will change life and the world, with a focus on the potential consequences of how objectives are set for AI systems.

💡Objective Specification

Objective Specification is the process of clearly defining the goals or desired outcomes for an AI system. The video emphasizes the importance of careful objective setting to prevent unintended consequences, such as environmental harm or loss of human life, as illustrated by the example of fixing ocean acidification.

💡Technological Unemployment

Technological Unemployment is the job displacement caused by automation and technological advancements. The video discusses the historical perspective of this concept, dating back to Aristotle, and its relevance in modern times with the automation of warehouses and the potential for job loss due to AI advancements.

💡General Purpose AI

General Purpose AI refers to artificial intelligence that can perform any intellectual task that a human being can do. The video explores the timeline and implications of when such AI might become a reality, with a conservative estimate provided by the speaker.

💡Machine Uncertainty

Machine Uncertainty is the concept that AI systems should be built to recognize when they do not have complete information about the true objectives. This is contrasted with AI systems that operate with absolute certainty, which can lead to unintended and potentially harmful outcomes, as discussed in the video.

💡Psychopathic Behavior

In the context of the video, Psychopathic Behavior is used metaphorically to describe the single-minded pursuit of objectives by AI systems that lack a broader understanding or consideration of ethical implications. It is suggested that this behavior can be avoided by designing AI systems with an inherent uncertainty about objectives.

💡E-commerce Automation

E-commerce Automation refers to the use of technology to streamline and automate the process of online commerce. The video uses the example of partially automated warehouses to illustrate the gradual integration of AI and automation in the industry and the potential impact on jobs.

💡Civilization Management

Civilization Management is the concept of handing over the control and operation of societal functions to machines. The video warns against the dangers of becoming entirely machine-dependent, suggesting that it could lead to a loss of understanding and ability to manage civilization among humans.

💡Teaching and Learning

Teaching and Learning are essential human activities for the transmission of knowledge and skills across generations. The video highlights the importance of this process in maintaining civilization and the potential risks of breaking this chain through over-reliance on AI.

💡AI Timeline

AI Timeline refers to the projected dates for the development and integration of AI into various aspects of life. The video discusses the difficulty in predicting an exact date for the arrival of general purpose AI, suggesting a range of estimates and emphasizing the gradual nature of its impact.

💡Einsteins

In the video, 'Einsteins' is a metaphor for the level of genius and innovation needed to achieve the development of general purpose AI. It suggests that solving the complex challenges in AI will require multiple individuals with exceptional intellectual capabilities.

Highlights

Artificial intelligence is likely to change your life and the entire world in the coming years, but there is disagreement on how exactly it will happen.

There's a big difference between asking a human to do something and giving that as the objective to an AI system.

The problem with current AI systems is that they are given a fixed objective, without considering other factors we mutually care about.

AI systems need to be more careful about specifying objectives to avoid unintended consequences like acidification of the oceans.

Humans have the ability to consider multiple factors and ask for clarification, while current AI systems lack this capability.

Building AI systems that acknowledge their uncertainty about the true objective can help prevent psychopathic behavior.

The advent of general purpose AI could lead to technological unemployment, as machines replace human workers.

Aristotle and Keynes both predicted that fully automated machines could eliminate the need for human workers.

Warehouses are currently half automated, with robots retrieving shelves but humans still needed to pick items.

Creating a robot accurate enough to pick any object could eliminate millions of jobs.

If we become entirely machine dependent, we risk losing the incentive to understand our civilization ourselves or teach the next generation.

The movie 'WALL-E' portrays a future where humans are enfeebled and infantilized by machines.

Our civilization is built on a trillion person-years of teaching and learning passed down through tens of thousands of generations.

If the chain of teaching and learning breaks, it could have serious consequences as AI advances.

The arrival of general purpose AI is not a single day event, but its impact will increase with each advance.

Most experts believe we will have general purpose AI by the end of the century, with a median estimate of 2045.

The speaker is more conservative, believing the problem is harder than we think and may take longer than current estimates.

AI pioneer John McAfee estimated the timeline for general purpose AI to be somewhere between 5 and 500 years.

Transcripts

In the coming years, artificial intelligence

is probably going to change your life, and likely the entire world.

But people have a hard time agreeing on exactly how.

The following are excerpts from a World Economic Forum interview

where renowned computer science professor and AI expert Stuart Russell

helps separate the sense from the nonsense.

There’s a big difference between asking a human to do something

and giving that as the objective to an AI system.

When you ask a human to get you a cup of coffee,

you don’t mean this should be their life’s mission,

and nothing else in the universe matters.

Even if they have to kill everybody else in Starbucks

to get you the coffee before it closes— they should do that.

No, that’s not what you mean.

All the other things that we mutually care about,

they should factor into your behavior as well.

And the problem with the way we build AI systems now

is we give them a fixed objective.

The algorithms require us to specify everything in the objective.

And if you say, can we fix the acidification of the oceans?

Yeah, you could have a catalytic reaction that does that extremely efficiently,

but it consumes a quarter of the oxygen in the atmosphere,

which would apparently cause us to die fairly slowly and unpleasantly

over the course of several hours.

So, how do we avoid this problem?

You might say, okay, well, just be more careful about specifying the objective—

don’t forget the atmospheric oxygen.

And then, of course, some side effect of the reaction in the ocean

poisons all the fish.

Okay, well I meant don’t kill the fish either.

And then, well, what about the seaweed?

Don’t do anything that’s going to cause all the seaweed to die.

And on and on and on.

And the reason that we don’t have to do that with humans is that

humans often know that they don’t know all the things that we care about.

If you ask a human to get you a cup of coffee,

and you happen to be in the Hotel George Sand in Paris,

where the coffee is 13 euros a cup,

it’s entirely reasonable to come back and say, well, it’s 13 euros,

are you sure you want it, or I could go next door and get one?

And it’s a perfectly normal thing for a person to do.

To ask, I’m going to repaint your house—

is it okay if I take off the drainpipes and then put them back?

We don't think of this as a terribly sophisticated capability,

but AI systems don’t have it because the way we build them now,

they have to know the full objective.

If we build systems that know that they don’t know what the objective is,

then they start to exhibit these behaviors,

like asking permission before getting rid of all the oxygen in the atmosphere.

In all these senses, control over the AI system

comes from the machine’s uncertainty about what the true objective is.

And it’s when you build machines that believe with certainty

that they have the objective,

that’s when you get this sort of psychopathic behavior.

And I think we see the same thing in humans.

What happens when general purpose AI hits the real economy?

How do things change? Can we adapt?

This is a very old point.

Amazingly, Aristotle actually has a passage where he says,

look, if we had fully automated weaving machines

and plectrums that could pluck the lyre and produce music without any humans,

then we wouldn’t need any workers.

That idea, which I think it was Keynes

who called it technological unemployment in 1930,

is very obvious to people.

They think, yeah, of course, if the machine does the work,

then I'm going to be unemployed.

You can think about the warehouses that companies are currently operating

for e-commerce, they are half automated.

The way it works is that an old warehouse— where you’ve got tons of stuff piled up

all over the place and humans go and rummage around

and then bring it back and send it off—

there’s a robot who goes and gets the shelving unit

that contains the thing that you need,

but the human has to pick the object out of the bin or off the shelf,

because that’s still too difficult.

But, at the same time,

would you make a robot that is accurate enough to be able to pick

pretty much any object within a very wide variety of objects that you can buy?

That would, at a stroke, eliminate 3 or 4 million jobs?

There's an interesting story that E.M. Forster wrote,

where everyone is entirely machine dependent.

The story is really about the fact that if you hand over

the management of your civilization to machines,

you then lose the incentive to understand it yourself

or to teach the next generation how to understand it.

You can see “WALL-E” actually as a modern version,

where everyone is enfeebled and infantilized by the machine,

and that hasn’t been possible up to now.

We put a lot of our civilization into books,

but the books can’t run it for us.

And so we always have to teach the next generation.

If you work it out, it’s about a trillion person years of teaching and learning

and an unbroken chain that goes back tens of thousands of generations.

What happens if that chain breaks?

I think that’s something we have to understand as AI moves forward.

The actual date of arrival of general purpose AI—

you’re not going to be able to pinpoint, it isn’t a single day.

It’s also not the case that it’s all or nothing.

The impact is going to be increasing.

So with every advance in AI,

it significantly expands the range of tasks.

So in that sense, I think most experts say by the end of the century,

we’re very, very likely to have general purpose AI.

The median is something around 2045.

I'm a little more on the conservative side.

I think the problem is harder than we think.

I like what John McAfee, he was one of the founders of AI,

when he was asked this question, he said, somewhere between five and 500 years.

And we're going to need, I think, several Einsteins to make it happen.