Recently, Anthropic's co-founder and CEO, Dario Amodei, posted a 10,000-word post that sparked a discussion on Twitter.
In this long article, Dario delved into how the future of AI can make the world a better place, and was praised by many as "one of the most rational and profound analyses of AI development in recent times".
At the beginning of the article, Dario Amodei made it clear that although he has been focusing on AI risks, this does not mean that he is a supporter of AI apocalypse, on the contrary, he is optimistic about the positive potential of AI.
"One of the main reasons I'm concerned about risk is that they're the only barrier between us and what I see as a fundamentally positive future. I think most people underestimate the huge benefits of AI, just as most people underestimate the severity of the risks. ”
So why do Dario and Anthropic talk about risk and safety, but rarely talk about the benefits of AI?
Dario says there are four reasons why:
- Maximizing leverage: Strong market forces can guarantee the benefits of AI development. Risks, on the other hand, are not predetermined, and our actions can dramatically alter the likelihood of them occurring.
- Avoid giving the impression of publicity: Many companies talk about the benefits of AI, which can easily distract the public and make people ignore the potential risks. And, "Spending too much time 'talking' is bad for your soul." ”
- Avoid pompousness: Many public figures and tech companies also like to exaggerate their mission when promoting AI, just as prophets lead their people to salvation. This religious view of practical technical goals is dangerous.
- Avoiding "sci-fi" baggage: Some of the current discussions about the future of AI are too "sci-fi" and full of unarticulated technical assumptions and cultural baggage, "reading like a narrow subcultural fantasy while offending most people." ”
In other words, "Render to God what is God's, and Caesar's to Caesar's." "The "secular" part of the economic benefits of AI should be left to the market; As practitioners who know more about AI, we should remind the public of the risks.
Dario's ideas are also reflected in Anthropic's dedication to transparency and controllability in AI.
There are reasons for this insistence: after all, Dario left OpenAI because he was skeptical of OpenAI's idea of accelerating commercialization, and Dario emphasized the need for stricter security when building robust AI systems.
Dario and several other former OpenAI employees also positioned Anthropic as a more security-focused alternative to OpenAI when they started their business, putting transparency and controllability first.
OpenAI 联合创始人 John Schulman、 Durk Kingma 和前安全主管 Jan Leike 也因为认同这份理念加入了 Anthropic。
Behind this security obsession is Dario's deep belief in the potential of artificial intelligence: the ability to squeeze the progress that humanity has made in the next 50-100 years into 5-10 years. This is what he calls the "compressed 21st century."
Dario defines AI in the next 5-10 years as powerful AI and believes that it will surpass Nobel laureates in its ability to excel in multiple fields such as math, programming, writing, etc.
Powerful AI will have a variety of human-computer interaction interfaces (text, audio, video, etc.), absorb information and generate results at a faster speed, and complete complex tasks autonomously. It can also control physical tools and equipment, and can even design its own robots.
Dario predicts that powerful AI could be mass-adopted by 2027.
When it comes to technological progress, some people hold two extreme views: one believes that technology will quickly trigger a "singularity" that will solve all problems at once, and the other believes that technological progress is saturated and that there is limited progress in intelligence superior to that of humans.
Thus, Dario borrowed from economics to propose the marginal effect of intelligence: when intelligence reaches a high level, other factors that complement intelligence become limiting factors. He argues that one should consider how other factors of production complement intelligence in a given situation and can be a limiting factor.
Within this framework, Dario envisions a blueprint for a future with powerful AI within 5-10 years.
The first is the field of biology, where many biologists have long been skeptical about the value of artificial intelligence and "big data" in biology. But Dario believes that AI is not just a tool for analyzing data, it can perform, guide and improve almost everything biologists do.
Dario points out that advances in biology often depend on a major discovery. If we could have more talented and creative researchers, these findings could occur 10 times more often or more.
For example, it was discovered in the 80s that CRISPR is a naturally occurring component of the bacterial immune system, but it took 25 years to realize that it could be used for general gene editing.
AI can speed up discovery by 10 times, but because experiments cannot be successful at the first time, and the speed of updating related hardware facilities is limited, AI cannot increase the speed of discovery by 100 times out of thin air.
Dario believes that in the future, artificial intelligence can help us prevent or treat most diseases, including infectious diseases, cancer, Alzheimer's disease, etc., and may also allow people to freely change their appearance, weight, etc., and greatly extend the life span of human beings.
Similar acceleration applies to neuroscience, mental health, economics, and more, but Dario warns that AI will not actively promote democracy and peace, but will require us as individual actors to drive the outcome.
At the end of the article, Dario talks about a classic question, "With AI doing everything, what's the point of being human?" How will humanity survive economically? ”
In response, Dario said he believes that meaning comes primarily from relationships and connections, rather than economic labor.
In the short term, there will be a healthy competition between AI and humans, and the excellent performance of AI will actually give rise to a new set of human jobs that complement and amplify the areas in which AI excels.
In the long run, the existing economic structure will be upended, and people may not need to participate in economic labor to obtain resources.
Leifeng.com
When is Opus 3.5 coming?
While discussing the content of the article, many netizens also asked when Opus 3.5 will be raised? Everyone waited for the flowers, thank you.
Very well written, so why is Opus 3.5 not released yet!
Based on past experience, the Claude collection is updated every three to four months. Claude 3 Haiku was released in March, and Claude 3.5 Sonnet went live at the end of June.
Dario himself says his plan is to push forward with front-line work every three months or so. Maybe by the end of the month we'll be able to use the latest Claude 3.5 Opus.
By the way, Dario thinks powerful AI like AGI could be as early as 2026.
However, Dario's vision of a future of AI and a large model like Claude are similar in form, and may be based on different architectures, may involve multiple interactive models, and may be trained in different ways.
This can't help but make people speculate, is it possible that Claude 3.5 Opus will make efforts in multimodal technology and launch image recognition functions similar to GPT-4? Is the delay in the rollout due to important improvements like AGI?
How AI is making the world a better place
Biology and Health
Biology is probably the area where scientific advances are most likely to directly and unequivocally improve the quality of human life.
In the last century, we have succeeded in eradicating ancient diseases such as smallpox, but there are still many diseases waiting to be conquered. Bioscience not only cures diseases, but also prolongs healthy lifespans, increases our control over our bodies, and solves seemingly unsolvable health problems.
However, the application of AI in biology faces challenges such as data, real-world speed, and inherent complexity. These factors are interrelated and slow progress in clinical trials.
For example, cell culture or chemical reactions can take days to weeks, while animal and human experiments can take months or years. In addition, the quality of the data is often low, and the causal relationships are ambiguous, making it difficult for scientists to understand biological phenomena.
The complexity of biological processes makes precise interventions difficult. Imagine a biochemical chart of human metabolism and you will understand the challenge. Coupled with the bureaucratic and regulatory requirements of clinical trials, progress is often delayed.
As a result, many biologists are skeptical about the effectiveness of AI and big data. While some progress has been made over the past 30 years, there have been only a handful of real breakthroughs.
For example, CRISPR was discovered back in the 80s, but it took 25 years to get applied. This suggests that there is insufficient support from the scientific community for promising directions.
In my opinion, the pessimistic view of AI is a misconception. I want to reiterate: the AI I'm talking about is used to perform, direct, and improve almost all of the work of biologists.
Advances in biology often stem from a handful of major breakthroughs, often associated with new technologies and measurement tools. Every year, there is about one major breakthrough that drives more than 50% of biological advances, enabling us to improve our understanding and control of biological processes.
I believe that powerful AI will dramatically accelerate the speed of these discoveries, allowing us to achieve 50 to 100 years of biological progress in 5 to 10 years. While factors such as experimental time and sequence dependence remain, large-scale parallel studies may advance this process. Despite the bureaucracy faced by clinical trials, rigorous drug evaluations have also accelerated progress. For example, COVID vaccines have been approved much faster than conventional medicines.
Ultimately, the track record of successful deployment in biomedicine is relatively strong. Despite the complexity of the drug development process, once successful, they are often used effectively, giving biomedicine a unique advantage in the field of technology.
To sum up, my base prediction is that AI-powered biology and medicine will allow us to squeeze the progress that human biologists have made in the next 50-100 years into 5-10 years. I call it the "compressed 21st century".
While it can be difficult to predict what powerful AI will do, we can look at the achievements of the 20th century to imagine possible future progress.
Here's what we might expect:
- Fighting infectious diseases: In the compressed 21st century, we are poised to effectively prevent and treat almost all 17 natural infectious diseases. With technologies such as mRNA vaccines, we may be able to eradicate infectious diseases, but of course, it also depends on the issue of poverty and inequality.
- Eliminate cancer: With cancer mortality declining by 2% per year, we could eliminate most cancers this century. Artificial intelligence will help us develop personalized treatment plans, and cancer morbidity and mortality are expected to be significantly reduced.
- Prevention of genetic diseases: With better embryo screening and CRISPR technology, we will be able to prevent and treat most genetic diseases, but interventions for systemic diseases may still be challenging.
- Breakthrough in Alzheimer's disease: Although the cause is complex, through artificial intelligence, we can hopefully find better treatments and even prevent the disease.
- Improving the treatment of other diseases: The treatment of diabetes, heart disease and other diseases will be greatly improved, and the mortality rate of many diseases has been significantly reduced.
- Achievement of biological freedom: Individuals will have greater control over their biological characteristics, such as weight and fertility, which will give everyone the opportunity to choose their own lifestyle.
- Increasing human lifespan: If we can find effective ways to slow down aging, life expectancy could increase from 75 to 150 years. This is not impossible, and there are drugs that can significantly extend the lifespan of some animals.
If these predictions come true in the next 7 to 12 years, the world will be very different. It would be a great victory for humanity to eradicate the disease that has plagued us for so long. The children of the future may never hear stories about cancer or Alzheimer's disease again, but will enjoy greater biological freedom and longer lifespans.
These changes will have a profound impact on our society, especially in social security and health care. If new technologies are widely available, the challenge in the future may turn to how to ensure that these benefits are available to all. While biology may be the only area where AI has successfully accelerated, even then the world could be turned upside down as a result.
Neuroscience and Psychology
In the previous section, I discussed physical illness and biology, but did not mention neuroscience and mental health.
In fact, neuroscience is an important part of biology, and mental health has a more direct impact on our lives. Hundreds of millions of people have a serious impact on their quality of life due to psychological problems such as addiction and depression, and billions more face mild psychological challenges.
Improving these issues will not only improve health, but also overall well-being.
The previously mentioned biological framework is equally applicable to neuroscience. Advances in this area often rely on key discoveries, such as optogenetics and CLARITY technology. AI has the potential to accelerate these advances, so the "5-10 years to 100 years progress" perspective also applies to neuroscience. In the 20th century, there were significant advances in neuroscience, such as in the 50s of the 20th century, when people understood how neurons discharge, so it is reasonable to expect artificial intelligence to advance neuroscience rapidly.
In addition, some knowledge of AI can help neuroscience.
Interpretability is an important example, and although biological and artificial neurons are different, they are similar in terms of the question of "how to calculate together". This explainability has recently been validated in mouse brains.
Experimenting on artificial neural networks is simpler than on real ones, making explainability an important tool for understanding neuroscience. Powerful AI may be able to develop and apply these tools more efficiently than humans.
At the same time, the training methods that AI teaches us could spark a revolution in neuroscience. Many people are concerned with learning issues that are not the same as I think they are. Simple objective functions and large amounts of data can drive complex behaviors, making us focus more on the goal and less on the details. Although I haven't paid close attention to this area, I feel like computational neuroscientists haven't fully taken these lessons yet.
I believe that neuroscientists should combine these fundamental insights with the uniqueness of the human brain to find the key challenges in neuroscience. While some of the issues are related to AI, AI neuroscientists may be able to use this idea more effectively to accelerate research.
I believe that AI will accelerate advances in neuroscience through four main pathways that will help heal mental illness and improve functioning:
- Traditional biology: AI will accelerate the development of drugs that modulate neurotransmitters, affect mood and cognition. It may also accelerate research into the genetic basis of mental illness.
- Meticulous neuromeasurement and intervention: Using techniques such as optogenetics and neural probes, we are able to measure and intervene in the activity of individual neurons, improving neuroscientific understanding.
- Advanced Computational Neuroscience: Insights from modern AI can be applied to the study of complex diseases, helping to uncover the root causes of psychosis and mood disorders.
- Behavioral interventions: Although I didn't go into depth in the discussion, there are many behavioral interventions in psychiatry and psychology, and AI has the potential to accelerate the development of these methods and help patients adhere to treatment.
My guess is that even without considering AI, the combined effect of these four pathways could cure or prevent most mental illnesses over the next 100 years, possibly reasonably accomplished within the 5-10 years of AI acceleration. Specifically, I argue that:
- Cure most psychiatric disorders: For example, disorders such as post-traumatic stress disorder (PTSD), depression, and addiction may be effectively treated through these four directions.
- Certain structural problems remain difficult: for example, differences in brain structure in people with psychosis may make certain conditions more difficult to treat.
- Effective genetic prevention: Mental illness is hereditary, and AI may help us prevent it through genetic screening, although this involves complex ethical issues.
- Improve everyday psychological problems: The psychological problems we face on a daily basis, such as anxiety and difficulty concentrating, may also be addressed.
- Elevate the human experience: Many people crave deeper creativity, compassion, and satisfaction, and AI has the potential to improve our cognitive and emotional abilities.
Although the concept of "mind uploading" is often mentioned, I believe that this idea faces huge technical and social challenges and is difficult to implement in the short term.
In conclusion, AI will dramatically improve mental health, enhance our cognitive and emotional abilities, and create a better and more humane future for humanity. This not only affects individuals, but may also improve many other seemingly unrelated issues in society.
Economic Development and Poverty
The first two sections discuss the potential of new technologies to cure diseases and improve the quality of human life, but the obvious question is: "Can everyone enjoy these technologies?" ”
Developing treatments for diseases is one aspect, but eradicating them globally is another. Currently, many health interventions are not yet available globally, and the same is true for technological improvements.
The standard of living in many areas remains poor. For example, sub-Saharan South Africa has a GDP per capita of about $2,000, while United States reaches $75,000.
It would be a serious moral failure if AI only helped developed countries and contributed little to developing countries. Ideally, therefore, strong AI should help developing countries catch up with developed countries.
I am not as confident in whether AI will be effective in addressing inequality as I am in the underlying technology for its invention. This is because economic problems often involve many complex human factors, and technological innovation can more easily overcome the barriers of complexity and data inadequacy.
Even in theory, AI cannot solve the "socialist computing problem", and I am skeptical that the government entrusts economic policy to such an entity.
In addition, there is also the question of how to get people to accept treatments that work but may be doubtful.
The situation is further complicated by the prevalence of corruption in developing countries. Corruption creates a vicious circle, with poverty exacerbating corruption, which in turn leads to more poverty. Therefore, AI-driven economic development plans need to take into account these human challenges.
Still, I remain optimistic. Historically, countries have succeeded in eradicating disease, and many places have moved from poverty to wealth, indicating the high intellectual rewards of these decisions. AI may be able to solve these problems more effectively than existing practices.
Targeted interventions may be able to circumvent human limitations, and AI should focus on these aspects.
Most importantly, we have to try. AI companies and policymakers in developed countries need to ensure that developing countries are able to participate, which is not only a moral imperative, but also a goal of our collective efforts. While success is not guaranteed, our actions can make a difference.
Five to 10 years after the advent of powerful AI, I have the following optimistic predictions for the future of developing countries:
Access to health interventions
In my opinion, the most promising area is the widespread distribution of health interventions around the world.
Many diseases have been eradicated through collective efforts, such as smallpox and polio, where cases have been reduced to very low levels. Artificial intelligence can play an important role in this process, enabling more efficient disease model analysis and optimization of logistics and distribution.
It is hoped that AI can help improve the efficiency of health charities and benefit more people. The development of new vaccines, such as single-shot vaccines against malaria, is simplifying the complexity of vaccine distribution.
In addition, the possibility of eradicating pathogens by releasing genetically modified mosquitoes is increasing. Overall, I expect developing countries to see significant improvements in health over the next 5 to 10 years, and possibly even surpass some developed countries.
Potential for economic growth
On the economic front, developing countries have the opportunity to quickly catch up with developed countries.
At the end of the last century, some countries in East Asia achieved an average annual GDP growth of 10%, successfully bridging the gap with developed countries. In the future, AI could help these countries create similar economic miracles.
By implementing effective economic policies and developing key industries, governments in developing countries can absorb the growth dividends of AI while respecting self-determination.
The implementation of many health interventions will be a major boost to productivity and, in turn, economic growth.
In addition, the application of AI in non-health sectors, such as more efficient energy and transportation technologies, will also open up new opportunities for developing countries. Despite the challenges, economic growth of 20% is an achievable goal.
Improvements in food security
In agriculture, advanced crop technologies, such as more efficient fertilizers and pesticides, will help increase crop yields. The application of genetic engineering can further promote this process, bring a second green revolution to developing countries, and alleviate hunger.
Tackling climate change
Developing countries are more vulnerable to the impacts of climate change, so technological advances in AI in the fight against climate change may help these countries find ways to reduce carbon emissions, such as clean energy and lab-grown meat. This not only alleviates environmental problems, but also contributes to economic development.
Inequality within countries
Inequality is likely to increase with the advent of advanced health technologies, but I am optimistic about the situation in developed countries. Market mechanisms often lower the barrier to entry for high-cost technologies, which, combined with the responsiveness of political institutions, can better drive technology adoption.
However, it is our shared responsibility to ensure that technology is inclusive.
Opt-out issues
One concern that alike in both developed and developing countries is that people opting out of the benefits of AI could end up in a bad feedback loop. But history shows that, despite the opposition, most people will eventually embrace beneficial technology. Therefore, we need to work to improve the scientific literacy of the public and make them understand the value of AI.
Overall, I am optimistic that AI can bring biological progress and economic growth to developing countries. Despite the challenges, we need to work together to move this process forward so that everyone can enjoy the dividends of technological development.
Peace and governance
Suppose all goes well: disease, poverty, and inequality are significantly reduced, and human living standards are dramatically improved. However, this does not mean that the pain is gone, and humanity still faces threats from each other.
While technological and economic development has helped to promote democracy and peace, such progress has been precarious and has often been subject to historical setbacks. For example, at the beginning of the 20th century, wars were thought to be a thing of the past, but two world wars followed.
Thirty years ago, Francis · Fukuyama predicted the eventual triumph of liberal democracy, but this ideal was not realized.
Twenty years ago, United States policymakers believed that free trade with China would drive China's liberalization, but the result was that we were challenged by a resurgent dictatorship.
Therefore, it is important to understand how AI affects peace, democracy and freedom.
Unfortunately, I don't believe that AI will automatically promote democracy and peace, although I think it has a positive role in health and poverty alleviation. Human conflict tends to be adversarial, and AI may serve both the "good guys" and the "bad guys". In particular, AI may enhance the capabilities of propaganda and surveillance, which are tools of dictators.
Therefore, as individuals, we must work to advance AI in the direction of promoting democracy and human rights. This is more urgent than focusing on international inequality: the triumph of liberal democracy is not taken for granted, but may require enormous effort and sacrifice.
This question can be divided into two parts: international conflicts and the internal structure of the state.
At the international level, it is crucial for democracies to dominate the global stage when powerful AI emerges.
AI-fueled authoritarianism is a cause for concern, so democracies need to put in place the conditions to channel powerful AI into the world to avoid being overwhelmed by authoritarianism and protect human rights.
Perhaps the best way to achieve this is through a "compact strategy," in which a coalition of democracies builds strength by protecting supply chains, accelerating development, and more. Such alliances could use AI to gain military superiority and distribute its benefits to more countries in exchange for support for their democratic strategies. The goal is to isolate the adversary and put it at a disadvantage in negotiations, thus giving up competition with democracies.
If this can be done, democracies will lead the way on the global stage, able to withstand the devastation of authoritarian states, and potentially turn the advantages of AI into lasting power.
This could lead to an "eternal 1991" and the fulfillment of Fukuyama's dream. But this requires private AI companies to work closely with democratic governments.
Even if all goes well, the struggle between democracy and autocracy within the country remains. Although it is difficult to predict the future, I am optimistic about the promotion of democracy by artificial intelligence. In a powerful AI environment, democratic governments can use their advantages to win information wars and counter dictatorships. Through global information channels, democracies can create a free information environment that limits authoritarian regimes' control over information.
Overall, improvements in the quality of life should promote democracy, and to some extent historically, it has been.
Improvements in mental health, well-being, and education are likely to boost democracy, as they are inversely correlated with support for dictatorship. After satisfying other needs, people want to express themselves more, and democracy is a form of self-expression. Authoritarianism, on the other hand, thrives on fear and hatred.
In addition, the flow of information helps to weaken the dictatorship. Uncensored AI may provide individuals with tools to help overthrow authoritarian regimes. Authoritarian governments often survive by depriving people of common sense, causing them to fail to realize the real problem.
We should also consider how AI can be used to make democracies better, not just prevent dictatorships. Despite the promise of equality in the law, there are injustices in practice. Can AI improve decision-making fairness? While there are concerns that it exacerbates discrimination, attention should also be paid to its potential to strengthen democracy.
The law is meant to be impartial, but it is usually interpreted by biased people. AI may be the key to solving this dilemma because it can make fuzzy judgments in a repeatable way. I suggest that AI be used as a decision-making aid, collaborating with humans, rather than replacing judges. Such a system needs to be transparent, allow people to understand its training process and assess possible biases.
In addition, AI can aggregate public opinion to help find consensus and resolve conflicts. Informed citizens are better able to strengthen democratic institutions. When it comes to government services, AI can help people access health benefits and social services, understand regulations in an easy-to-understand way, and enhance respect for democratic governance.
While these ideas may seem vague, we should pursue the vision of AI as a guarantor of freedom, individual rights, and legal equality. In the 21st century, AI-enabled polities can be powerful protectors of individual freedoms and beacons of the world's desired form of government.
Work & Meaning
Even as we make progress in reducing disease, poverty, and inequality, liberal democracy becomes mainstream, and existing democracies become better, we still face an important question: What is the point of human existence when AI is so advanced? How will they survive?
This problem is more complex than others, mainly because it involves macro issues of social organization that can often only be solved naturally over time. Hunter-gatherer societies throughout history may have felt that life without hunting and rituals was meaningless, and that there was a lack of understanding of our society today.
Nevertheless, I would like to touch on this issue. It may not be correct to think that AI does a better job that means that human tasks are meaningless. Most people aren't the best in any field, but that doesn't stop them from finding value.
People like to engage in activities that don't create economic value, such as playing games, swimming, or partying with friends.
Meaning comes more from human relationships than from mere economic labor. In the post-AI world, the sense of accomplishment and competition in the pursuit of complex tasks remains.
However, the question about the economy is more intractable. The challenge we may face is that most people may not be able to contribute meaningfully to an economy that is highly dependent on AI in the future.
While comparative advantage may help humans stay relevant and be more productive in the short term, the widespread use of AI could change the landscape in the long term.
Historically, civilizations have successfully undergone many economic transformations, from hunter-gatherers to agriculture to industry. In the future we may need new solutions that are not foreseeable today and may include methods such as universal basic income.
Perhaps the economic model will rely on the judgment of artificial intelligence to give humans the resources they need, although these ideas have yet to be verified.
In the face of these challenges, we need to explore, experiment, and work to avoid potential exploitation and dystopian situations. More discussion and reflection will be needed in the future to find a suitable way out.
summary
In the previous discussion, I painted a picture of a bright future: if AI goes well, the world will be a better place than it is now. However, it will take a lot of brave people to work together to achieve this vision. Everyone, including AI companies, needs to be actively involved to prevent risks and share in the benefits.
It's a goal worth striving for.
If, in the next five to 10 years, we can defeat most diseases, advance human freedom and democracy, and lift billions of people out of poverty, everyone will be shocked by the change. We will not only be amazed by the benefits of new technologies, but also moved by the gradual realization of long-held ideals.
In the process of writing, I realized that this vision is quite radical and may have struck many as absurd, and some even disagreed with the values and political choices in it.
But this vision also presents a fascinating possibility, as if different visions of a better world are moving in the same direction.
In Iain M. Banks' Culture series, the main character comes from a society with similar principles to what I described. They are in an authoritarian empire where they decide their leadership through complex battle games. The protagonist's victory over the emperor with cultural values shows that even in a competitive environment, cooperation and empathy are the keys to victory.
Scott Alexander also mentioned that competition often leads to a society based on compassion and cooperation.
I believe that the values of culture are a successful strategy. These values, which stem from a shared understanding of fairness, cooperation and autonomy, make it easier for us to build a force of unity. Everyone agrees that children should not die of illness and that every child should have equal rights. On this basis, we can work together to achieve this goal with wisdom.
There is general agreement that it is reasonable to punish those who unnecessarily attack or harm others, and that this consistency should be reflected in laws and institutions. At the same time, the autonomy of the individual over his or her own life and choices is evident.
Ultimately, the logical derivation of these intuitions will lead us to the rule of law, democracy, and Enlightenment values. The advent of artificial intelligence provides us with an opportunity to achieve these goals and a clear direction for the future.
Still, it's a transcendent beauty. We have an opportunity to play a small role in making this happen.
Leifeng.com