Amazon founder Bezos in depth interview ChatGPT is not an invention, but a discovery

Visual China.

Titanium**Note: This article ** on WeChat *** love fan er (id: ifanr), author |Fang Jiawen, Titanium** is authorized to publish. Since stepping down as Amazon's CEO in 2021, Jeff Bezos has rarely been seen in the public eye.

Recently, Bezos chatted for two hours on the renowned tech podcast Lex Fridman Podcast, sharing his work at Blue Origin, insights on generative AI, and more on management

Lex Fridman and Jeff Bezos

About AI: Bezos feels that today's large language models are more like discoveries than inventions, because they are not fully controllable and often surprising. Sharing Blue Origin's inaugural space voyage experience: he seemed to witness his own funeral;The feeling of weightlessness is wonderfully natural;It's astounding to look back at the earth. Blue Origin's mission is to be the most decisive company in the world, and Bezos shares how to build this decisiveness and why truth-seeking is an eternal pursuit. Meeting Aesthetics: Bezos on why he resists PPTs and how to use a six-page memo to have an effective meeting. The Future of Humanity: Bezos talks about the possibility of human life in space in the future, and trillions of people are not impossible. For generative AI like ChatGPT, Bezos comes up with a rather interesting definition: Large language models in their current form aren't inventions, they're discoveries. 」

In Bezos's view, only something that we deliberately design and have a clear understanding of how it works is an invention.

For example, the telescope was invented, but seeing Jupiter through a telescope and knowing that it has its own moons is a discovery.

Large language models are more like discovery. We are often amazed at what they are capable of. They are not designed to be.

As for the potential harm of AI to human survival, Bezos shows optimism:

There are many ways in which we humans can perish ourselves.

These technologies may help us not to do these things, but they will save us.

Subsequently, he extended the discussion with the "Oppenheimer" movie, but the person he focused on was Louis Strauss played by Robert Downey Jr.:

We've invented these amazing, devastating technologies, called nuclear, and they're managed by us humans, but we don't really have the power to control them.

And [Strauss] he represents all this ...... in this filmHe was stingy there, thinking Oppenheimer had spoken ill of him to Einstein.

We, as a species, haven't really developed advanced and mature enough to manage these technologies.

Compared to these models, we know that the human brain does work differently, in part because we are very energy-efficient.

The human brain can do amazing things, and it only takes about 20 watts to do them. AI is much more energy-intensive to do the same thing.

Autonomous driving technology requires tens of billions of miles to learn to drive, but humans don't need to drive so many miles to learn it.

This also inspires us to develop AI:

So, there are still some tricks that we haven't learned yet.

I think the key is not just about the scale [of the training data].

Interestingly, we are now paying off by simply scaling up our data.

How fast [AI] can grow, no one knows. But in the next decade or twenty, I think we're going to see very significant progress.

I'm personally excited about it.

What everyone didn't know was that just before the launch, Bezos seemed to have seen a funeral of his own.

At 4:30 a.m. that morning, Bezos and his brother were preparing to leave for the launch site when they saw their family and good friends seeing them farewell:

They say goodbye to us in a way that they feel like they might be saying goodbye to us......You can see how nervous they are.

It was a very impactful scene because it was almost like attending your own funeral and you could feel how much you were loved, and it was really great.

Bezos said that when he was going to tell his mother that he was going to take his brother with him on the maiden flight, it would be difficult

She reacted similarly, both of you (both go)?」

Four passengers on Blue Origin's first manned space journey.

Contrary to saying goodbye to relatives and friends, Bezos was relaxed.

Bezos himself is very familiar with the New Shepard, and, as a tourism project that will be available to the general public in the future, Bezos knows how much effort has been put into safety:

The amount of time we spend on the escape system is the sum of the time we spend on the rest of the rocket. It was the toughest part of the entire New Shepard architecture.

In his opinion, no matter how the design is optimized, there is no guarantee that a rocket launch will never go wrong, so the only way to improve safety is to set up an escape system.

Once the launch began, Bezos enjoyed it straight away.

One thing that's particularly interesting is that the zero-gravity state feels particularly natural.

I don't know if it's like going back to the womb or whatever.

You can see how fragile the planet is. If you weren't an environmentalist in the first place, that moment turns you into an environmentalist.

The great Jim Lovell (astronaut) once said that it was when he looked back at the Earth from space that he realized: You don't go to heaven after you die. You came to heaven when you were born.

In space) you see the infinite darkness, and there is a jewel of life in all nothingness, and that is the earth.

Blue Origin needs to evolve faster.

The main reason I'm doing this is so that I can spend a little more time at Blue Origin, put in some more energy, add some sense of urgency, and we need to move faster than we do now.

In the past, he set Amazon a mission to be the most customer-centric company in the world—starting with customer needs and working backwards.

For the world's most decisive companies, Bezos says it means that we're going to be very good at taking technically appropriate risks and making those decisions quickly.

At the same time, decisiveness does not mean that it cannot be reversed :

We can always change our minds.

I often talk about one-way doors and two-way doors, and most of the decisions are two-way doors.

The one-way door refers to a very important, irreversible decision, and when you step through that door, there is no turning back.

Bezos will be very cautious when it comes to one-way door decisions, thinking in many ways, because we really have to get it right from the start.

He jokes that he was the Chief Slow Down Officer at Amazon because he had to make decisions about slowing down one-way door decisions.

In his opinion, many large companies are slow to move because they are treated as one-way doors, whether they are one-way or two-way doors

Most two-way door decisions should be made by an individual or a small team deeply rooted in the company.

One-way doors should be slowed down by top management to ensure that the right choice is made.

This means that even if Bezos can't agree with his subordinates, he thinks that the other party may understand things better and trust them, and he will directly say that he doesn't agree with this point of view, but is willing to support and help him try

I'm going to take the initiative to help them do that. This is very important teammate behavior.

Lex Fridman visits Blue Origin.

As a counter-lesson to the counterpart, he singled out two particularly bad ways to make decisions: compromise and fighting who would make it to the end.

Compromise, it has the advantage of saving trouble, but it does not lead us to the truth.

The correct thing to do is to use a ruler to measure. But it takes time and effort to get a ruler and an idea to measure.

Many times, no one knows what the true truth is, and this is where disagree, but put your heart and soul into it.

Escalate the problem to your boss and tell him: We can't get to a consensus. We like each other. We also respect each other, but we can't agree with each other on this issue. We need you to make a decision before we can move forward. 」

Moving fast on decision-making, as quickly as possible responsibly, that's how you get up to speed.

Meetings are his daily work.

In my mind, the perfect meeting starts with clear documentation.

For the first 30 minutes of each session, all participants quietly read the six-page memo together.

Why don't you let someone else read it in advance?Bezos said that people don't always have time to read early, which makes it difficult for them to keep up in meetings, and even pretends to have read it like they did in college, which doesn't help solve the problem.

Make that time for everyone to start from the same starting point of information.

This also ensures that the proponents can express their true thoughts without being influenced by their leaders.

Also, Bezos doesn't like PPTs.

PPT is designed to be a persuasive tool. It's like a sales tool. And internally, the last thing you want is to be sold.

Again, you are looking for the truth.

Although the half-hour plan for the first half hour of the meeting is super fixed, when it comes to the real discussion part of the meeting, Bezos will be very free because he believes in the importance of mind wandering.

In his view, the core of the meeting is to ask questions that no one knows the answer to, and everyone is looking for solutions through divergent thinking.

When everything is right, it makes all the time spent in other meetings worthwhile.

That feels good. When you get a real breakthrough in a meeting, it feels like there's a beauty in it.

If we had a trillion people, we would have 1,000 Mozart and 1,000 Einstein at any given time. Our solar system will be teeming with life, intelligence, and energy.

However, to accommodate so many people, it is a bit difficult to live on a planet, and the form of a space station may be more suitable.

We can take resources from places like the Moon, NEOs, and asteroids to build huge O'Neill colonies on which people will live.

The O'Neill colony consists of two counter-rotating cylinders, each about 5 kilometers in diameter and 20 kilometers long, whose rotation provides artificial gravity to the O'Neill cylinder.

O'Neill colony concept depicted.

In Bezos's view, humans still want to live closer to Earth after moving to space, especially in the early days.

Humans can choose to live in space or on Earth, but the difference is that the entire human race can use resources from space.

Humans who have moved there may still want to go back to Earth for a vacation, just as people want to go to Yellowstone now.

For Bezos, the migration of humans to space is inevitable.

He believes that although people like to talk about the good old days, modern society is indeed better than before, and the poverty rate, mortality rate, etc. have been greatly reduced.

The only thing that has deteriorated is the natural environment.

We have traded some of the beauty of pristine nature for other benefits of our modern society.

We can have both, but in order to get there, we have to go to space.

We have sent robots to explore all the planets, and we know that this planet (Earth) is really good.