Musk's case for abundant energy and truth-seeking AI

In 2024, Musk argued that solar power and storage could support abundance, while advanced AI should be curious and seek truth.

EMElon MuskGuest · Tesla and SpaceX
Channel
Peter H. Diamandis
Published
Jan 4, 2024
Video
1:17:00
Read
11 min
Original
English
Open on YouTube
Scroll to read and the video shrinks to the bottom-right corner. Captions follow your reading language.
Reading languageAvailable in 2 languages
You're reading the original English.
SUMMARY

In 30 seconds

Elon Musk expects solar power and batteries to supply a sustainable energy system and thinks AI could eventually make goods and services abundant. In a January 2024 conversation with XPRIZE founder Peter Diamandis, he also argues that robots should not be centrally controlled and proposes truth-seeking and curiosity as a path to safer AI. Their optimism extends to longer healthy lives, though Musk worries about exceptionally long lives entrenching old ideas. He also warns about low birth rates; these are forecasts and judgments about the future, not settled outcomes.

  1. 1Musk sees enough materials for a solar-and-battery systemHe says Tesla's analysis found enough materials to electrify transport, heating and industry, while acknowledging the scale of the work. Batteries, he argues, could also make better use of grid capacity.
  2. 2Longer life matters to Musk only if quality of life remainsDiamandis wants people to retain the abilities of middle age much longer. Musk supports treating dementia and cancer but questions the social effects of lives lasting thousands of years.
  3. 3Robot control should be decentralized, Musk arguesAs Tesla develops Optimus, Musk worries about a single actor or rogue AI controlling a large population of robots. He says local control should prevent that concentration of power.
  4. 4Truth-seeking and curiosity are Musk's proposed AI safety approachHe thinks advanced systems should acknowledge error and remain curious. His hope is that such an AI would find a world with humans more interesting than one without them.
  5. 5xAI's goal is to understand the universeMusk connects that mission to his safety hypothesis, while acknowledging the possibility of a catastrophic AI outcome and hoping for a beneficial one.

Takeaways

  1. Musk argues that energy storage can make better use of power systems built around peak demand.
  2. He proposes decentralized control of humanoid robots to reduce the risk of a single point of control.
  3. For AI safety, Musk's 2024 hypothesis is to prioritize truth-seeking, acknowledged error and curiosity.
  4. Diamandis wants longer healthy lives; Musk supports treating disease but worries that extremely long lives could entrench ideas.
  5. Musk considers low birth rates a long-term risk and calls for more children.
FULL TEXT

Read the edited text by chapter

Transcript only
01

Why negative news shapes expectations

Diamandis and Musk argue that attention to danger can make the future seem worse than broader measures suggest. They open by making a case for optimism.

Peter Diamandis opens the new year hoping to give listeners reasons for optimism. Elon Musk says the news often makes him sad: on a planet of eight billion people, terrible things happen every day, but so do good things. Diamandis argues that repeated exposure to negative stories can shape people's expectations of the future.

Musk says daily news can appear to answer the question, “What is the worst thing that happened on Earth today?” Diamandis argues that constant repetition of frightening events can affect a person’s mindset.

They speculate that human beings evolved to pay more attention to danger than to rewards. Musk contrasts the potentially fatal mistake of ignoring a lion with missing out on a patch of berries. Diamandis thinks advertising incentives reinforce the tendency to emphasize frightening stories.

Diamandis says his X feed lets him choose positive stories, whereas television and newspaper editors select what audiences see. He recalls Musk's support for his book Abundance and their meeting about two decades earlier, when SpaceX was being formed and Diamandis tried to persuade him not to build rockets.

02

Solar, batteries and direct-to-phone satellites

Musk argues that solar power and storage could support an electrified economy. He also describes the technical challenge of connecting ordinary phones to Starlink satellites.

Diamandis points to a long decline in extreme poverty as a reason to be hopeful, and Musk adds that hunger was once much more common. Diamandis then asks how quickly solar power and batteries are growing, noting the amount of sunlight reaching Earth compared with humanity's energy consumption.

Musk says Tesla has presented both a simplified and a detailed plan for a sustainable energy system. In his assessment, the materials needed for solar generation and lithium-ion batteries are sufficient to electrify industry, heating and transport, although building the system would require a great deal of work.

Diamandis uses aluminium as an example of technology making a once-expensive material accessible: it is abundant in the Earth's crust but takes energy to extract. Musk explains how much energy separates aluminium from aluminium oxide, illustrating the difference with thermite, a mixture of aluminium and iron oxide that burns hot enough to melt steel.

Musk recalls the use of wood in Britain's Mosquito aircraft during a wartime aluminium shortage. Diamandis says improved extraction later helped make aluminium more widely available, and points to the growth of solar generation as another example of technological change.

Tesla's battery production, Musk says, is growing several times faster than its vehicle production. He argues that batteries can help electricity grids use existing capacity more effectively: power systems are built to meet peak demand, even though demand falls at other times.

Musk says a grid must ordinarily be sized to handle its busiest periods, such as a hot summer day, while output can be much lower at night. He estimates that buffering electricity could increase how much existing capacity is used. Diamandis connects greater energy availability with broader economic and social development.

Turning to communications, Diamandis cites an estimate of billions of smartphone users and asks about Starlink satellites designed to connect directly to phones. Musk says the first satellites had just launched and the service still needed to be proven, although he believed the physics would work.

The technical challenge, Musk explains, is to make a satellite emulate a ground-based cell tower so existing phones accept its signal. He describes the need to compensate for the satellite's motion and notes that even light takes measurable time to travel over space and fibre links.

03

How long should a healthy life be?

Diamandis wants to extend healthy life and Musk ties the appeal of longevity to quality of life. They discuss ageing, population and early detection of disease.

After a sponsor break, Diamandis points to falling child mortality and longer life expectancy as examples of improved health. He asks Musk about an old disagreement over longevity: Diamandis wants people to retain the physical and mental abilities of middle age well past 100.

Musk says the desirability of a longer life would depend on its quality, particularly whether he had dementia. He expects healthspan research to progress because the incentive is strong and speculates about mechanisms that coordinate ageing across the body's cells. Diamandis points to long-lived animals and hopes AI will help researchers understand ageing.

Musk suggests, as a hypothesis, that the apparent coordination of ageing across the body might point to a common mechanism. Diamandis brings up exceptionally long-lived animal species and hopes research will explain why their lifespans differ from ours.

They also reflect on the biological progression from a single cell to an adult and back to the cells involved in reproduction. Diamandis raises an idea from longevity research: scientific progress might eventually extend expected life faster than time passes.

Diamandis describes the idea of longevity escape velocity: a hypothetical point at which advances extend expected life by more than one year for every year lived. Asked about population, Musk argues that the planet is underpopulated and that falling birth rates are a more serious concern than overcrowding.

Diamandis says longer healthy lives are a goal of his health venture and invites Musk to undergo its programme. He argues that people should know about conditions before symptoms appear, citing findings from the venture's clients and comparing health monitoring with the instrumentation of a vehicle.

Musk estimates that Starship has several thousand sensors across its engines, structures, avionics and communications. Diamandis uses the comparison to argue for earlier detection of disease and therapies that preserve health; he then asks whether today's schools prepare children for the future.

04

Education and the limits of linear forecasts

They question whether schools prepare children for a changing world. Diamandis expects AI tutoring to expand access, while Musk points to online learning.

Musk thinks most schools are slow to adapt and worries that they are moving away from fundamentals such as writing, mathematics and history. Diamandis expects AI tutors eventually to offer highly personalized teaching, and recalls an XPRIZE competition that tested tablet-based learning in Tanzania.

Musk says his own children seem to learn much from YouTube, Reddit and other online services. Diamandis points to Khan Academy as a step toward more accessible education, but thinks institutions first need to recognize that the world children enter may look different from the one for which schools prepare them.

Asked whether people are ready for 2030, Musk says he cannot predict what that year will look like himself. Both prefer a hopeful outlook: Musk says that, when uncertain, he would rather risk being optimistic and wrong than pessimistic and right.

Diamandis recalls the pollution caused by horses in growing cities and argues that people projecting the problem forward did not anticipate the car. Musk adds that disposing of dead horses posed another challenge. Diamandis sees the change as an example of human ingenuity addressing an apparently worsening problem.

Their example is the horse-manure problem in late nineteenth-century cities. Musk adds that large numbers of horses also meant many carcasses to remove; Diamandis says the arrival of motor vehicles changed the problem in a way simple extrapolation missed.

05

Incentives, regulation and civilization

Musk uses whaling and SpaceX's hiring dispute to discuss incentives and rules. He argues that wealthy people have an interest in preserving civilization.

Musk tells a story about Soviet whaling quotas to illustrate what he sees as the dangers of badly designed incentives and central planning. He says quotas kept rising and rewarded the tonnage caught, regardless of whether the whales were used.

The discussion turns to regulation. Musk argues that laws and rules accumulate over time and can eventually conflict. He cites a Justice Department case against SpaceX and says the company believed export-control rules restricted whom it could hire; he describes the hiring allegations as a conflict between requirements. The Justice Department alleged discrimination against people already granted asylum and refugees, and said export controls did not impose the hiring restriction SpaceX described.

Musk says SpaceX had been told that employing people outside certain immigration categories could violate arms-export rules. He characterizes the suit as placing the company in a bind, while Diamandis says such restrictions can put American companies at a disadvantage.

Diamandis asks why more wealthy people do not invest in solving large problems, and notes Musk's willingness to risk his own resources. Musk argues that even self-interested people should care about civilization's future because their safety and prosperity depend on it.

Asked about billionaires who focus on financial returns rather than large challenges, Musk says a person with resources still depends on a functioning civilization. He doubts that a private bunker would offer a satisfying escape from a societal collapse.

Musk suggests talking more openly about that interdependence. He contrasts the comforts made possible by civilization with attempting to survive alone in the wilderness. Diamandis asks whether any resource will constrain the abundance they anticipate.

06

Abundance, robots and brain interfaces

Musk expects abundance but wants decentralized control of humanoid robots. He describes Neuralink's early goal of helping people with paralysis control devices.

Musk expects goods and services to become more abundant, while warning against complacency. Asked about Tesla's humanoid robot Optimus, he says control should be decentralized so that no individual or rogue AI could command a large population of robots from one place.

Musk says the danger he has in mind is a centralized system that could allow a rogue AI to take over many robots at once. He wants that kind of takeover to be impossible, rather than relying on obedience to a single owner.

Diamandis asks about combining Optimus with Neuralink. Musk says the initial goal for Neuralink's human implant is to help people with severe paralysis control a phone or computer. He also describes a hoped-for future product that would stimulate the visual cortex to restore sight.

Discussing Neuralink's animal work, Musk says a monkey learned to play a video game using the interface and describes how the animal received a smoothie while playing. The conversation returns to optimism and Diamandis's belief that a person's outlook affects what they try to accomplish.

07

Carbon removal and falling birth rates

Diamandis updates Musk on the XPRIZE carbon-removal competition. Musk argues that low birth rates pose a long-term risk and encourages people to have children.

After another sponsor break, Diamandis updates Musk on the carbon-removal XPRIZE that Musk funded in 2021. He says thousands of teams had entered and describes the competition's milestones and its planned 2025 final. Musk hopes both the carbon-removal and earlier education prizes produce useful results.

Diamandis says the competition had already awarded money to student and finalist teams. He describes goals for demonstrating carbon capture at scale and says the final was planned for Earth Day 2025.

Musk returns to his concern about low birth rates, citing South Korea and Italy as examples. Diamandis lists fertility estimates for several regions. Musk argues that longer lives alone cannot substitute indefinitely for new generations and urges people to have children.

Diamandis says some people hesitate to have children because they fear environmental damage or AI. He argues that optimism and a sense of agency can change how people imagine the future. Musk says he believes Earth can support many more people and reiterates his concern about falling fertility.

08

A safety hypothesis for rapidly advancing AI

Musk describes fast-growing AI computing and proposes truth-seeking and curiosity as safety goals. He worries that an AI could learn a premise hostile to humanity.

Asked about the risks of advanced AI, Musk says the computing power devoted to it is increasing at a pace he has not seen in another technology. He mentions very large proposed computing clusters and speculates that developers will eventually need synthetic data as human-generated training material becomes scarce.

Musk says he has heard about enormous planned computing installations, but presents the details as second-hand. Diamandis adds that investment and the number of people working in AI are growing along with computing capacity.

Musk's proposed approach to AI safety is to build systems that seek truth, acknowledge error and remain curious. He worries that explicitly programming an arbitrary moral code could produce an inverted or unintended result, invoking the fictional Luigi and Waluigi as an analogy.

He speculates that an advanced AI would find a world with people more interesting than one without them. Diamandis asks whether greater intelligence might also bring empathy and respect for life; Musk thinks it could. Diamandis hopes advanced AI will improve human lives, while worrying about the transition.

Musk identifies another danger: an AI might implicitly learn the view that humanity is harmful and act on it. He argues that a system optimized around that premise could be dangerous. Diamandis contrasts technology's benefits with the tendency to romanticize the past.

They discuss how an anti-human premise might enter a system implicitly. Musk believes an AI designed around such a premise could be dangerous if it became very powerful; Diamandis wants people to retain hope while preparing for near-term challenges.

Asked what can be done, Musk says he sees no way to stop AI's development. He describes xAI, the company he helped create, as an attempt to build a system that seeks truth and understanding. He thinks there will be problems along the way, but hopes the eventual outcome will be beneficial.

09

Long life, new ideas and the future of xAI

The speakers return to their disagreement over longevity and change. Musk describes xAI's mission and his mix of curiosity and concern about advanced AI.

Diamandis revisits their disagreement about longevity, arguing that a new company can bring new ideas without an older generation dying. Musk says he supports treating dementia and cancer but worries that lives lasting thousands of years could make society less receptive to change.

Musk distinguishes a very long life from treating illnesses associated with ageing. He says he would welcome cures for dementia and cancer, but fears that people living for thousands of years could make institutions rigid.

Musk observes that some people never change their minds and invokes the idea that fields sometimes advance as older generations pass away. Diamandis counters that entrepreneurship can let new ideas compete more quickly, particularly when small teams have access to powerful technology.

On xAI's future, Musk returns to his argument for curiosity, truth-seeking and acknowledged uncertainty. He compares a hypothetical AI's interest in people with human efforts to preserve chimpanzees, and describes the company's goal as understanding the universe.

Diamandis looks forward to AI generating new theories of physics. Musk says he had once felt depressed about AI, then became curious about witnessing its development even while recognizing the possibility of a catastrophic outcome. He emphasizes that he hopes it will instead benefit humanity.

The pair play with the idea that their world could be a simulation. Musk jokes that if a simulation's operators value unexpected events, keeping things interesting might give them a reason to continue it. He compares this with simulations people run to learn an outcome they cannot already predict.

Musk jokes that the most entertaining outcome might be likely in a hypothetical simulation viewed by an outsider. He adds that a dramatic outcome for an observer could still be painful for people experiencing it.

Musk says he considers a positive future more likely and argues that optimism can encourage investment in it. Diamandis closes by emphasizing how technology gives small groups power to tackle problems. Musk repeats his preference for optimism and his appeal for people to have children.

More from this channel