The idea of general AI has long captured the imagination of people across the world. But what would society and the economy look like in a world with general AI? And has the response to the coronavirus given us an insight into some possible trends and changes?
By general AI or AGI (also called strong or human-level AI), we mean an AI which has the ability to learn and solve new problems on its own. One can imagine it as an AI scientist that generates specialized AIs capable of solving novel problems. This is very different from the AI available at the moment.
There has been a lot written about the existential risks of AI and many apocalyptic style scenarios have been appearing in popular culture. However, there is a flip side to the coin. In this blogpost we imagine a future where this technology is used for good to mold our society for the better and help humans reach their full potential.
1. Helping others
According to many schools of thought, but also according to increasing research on happiness, it is our relationships with others which is one of the most important determinants of our own happiness (together with good health). If general AI can be used to create an abundance it may allow us to focus less on material needs and more on the people around us; this includes helping not just our friends and family but our communities. By fostering this human interaction we are likely to see a happier and more inclusive society.
2. Increasing freedom
AI could help better analyze the future consequences of proposed public policies and allow citizens to have informed deliberation and potentially direct vote on the most important issues. Political parties, think-tanks, and civil society organizations would serve as catalysts of these fact-based deliberations and media reporting could become significantly more fact-based. All these stakeholders would compete to see who could deploy AI better to increase overall social well-being and individual freedoms. In democratic liberal societies, the main imperative for future regulation of AI should be ensuring that individual freedom and autonomy is not being reduced by certain deployment of AI.
However, there are various related dangers. In particular, we need to be aware of the potential of AI to be abused for various types of manipulation, and we need to establish mechanisms to prevent such abuse. For example, Software Engineer and AI Researcher François Chollet has outlined one such example as the “echo chamber,” created by recommender systems. Chollet argues that an adversary, or a government, could use machine learning to shape people’s opinions by supporting or suppressing the voice of an individual, simply by exposing them to audiences with similar or opposing opinions and thus generating an impression of confirmation or rejection. Even without explicitly bad intentions, such trends would be alarming, and as Chollet points out we must become the masters of AI algorithms and not the servants. This is where AGI could potentially help, if this technology is not monopolized by a few actors but becomes accessible to the people, giving them better tools for the analysis and filtering of the information and ability to protect themselves against negative practices.
3. Protecting privacy
An area of great concern for AI development has been the privacy of individuals. With an increasing move into digital, and even more with the personalization of products and services, comes the generation, processing, and storing of large amounts of very detailed personal information. The coronavirus has brought this issue to the forefront. Various governments use advanced surveillance tools, in order to predict infected people and to track their movements. These tools have been effective in controlling the disease and are justified in times of emergency, however, privacy protection advocates, as well as many thinkers such as Yuval Noah Harari, fear that they will continue to be used and become a common part of life after the crisis. It is likely that this trend would have occurred anyway, however the crisis seems to have sped it up. This has prompted various groups to work on solutions to how we can use the technology while respecting people’s privacy. While AI has a huge potential to erode people’s privacy, it also has the opportunity to safeguard it. Various privacy-protecting tools could be significantly enhanced effectively creating an AI-personal-aid which will ensure only the information an individual wants to share is shared.
4. Realizing the potential of every single individual
Once advanced AI eases the economic stress and maximizes efficiency, people will be able to invest more in making their lives better. Employers could offer the opportunity for self-development to each individual who would then have the freedom to choose how they want to develop. Regular sabbaticals could become the norm, giving employees the chance to explore other fields or take time to explore their creative tendencies. Governments or private organizations could establish stipends to motivate and support individuals to learn new things and improve. Stipends could also be available to all college students, allowing everyone the chance to study without stress. As well as investments into people, there would be more capital freed up to invest into well-being related research and development as well as the social sciences, which could lead to not just a more technologically advanced society, but in a broader sense culturally more advanced, and a happier one too.
The future of education is unlikely to look anything like it looks now. With general AI the potential to create truly individually tailored education for each pupil would be possible, allowing each student to get the attention they deserve and need. The virtual AI teacher will be caring, empathetic and patient. The teacher can also take on the image of a friendly character. The AI also has the potential to reduce class sizes where it is beneficial and help create high-quality education throughout a child’s life, starting in pre-school.
5. Humanizing essential services
The invention of general AI would also allow for the qualitatively new improvement and personalization of products, but perhaps more importantly, services. For example, AI will be able to take a large workload off of doctors by taking on work such as identifying health issues in a holistic fashion, leveraging the knowledge across medical fields, and personalizing treatment and medication using patient’s individual data. These will not only include individual genome data but also other data such as information about people’s surrounding environment and variables which will help to forecast the expression of various genes and make possible very early preventive changes. In general, AI will also take some strain off the medical facilities and help reduce waiting times.
AI could help automate the mundane tasks undertaken by medical personnel or carers, like delivering food or monitoring the patients, and allow them to dedicate more of their time to caring, conversation, and other very humane interactions. In his book, AI Superpowers, Kai-Fu Lee emphasizes the importance of focussing on personal interactions where humans are likely to still hold an upper hand over AI.
6. Healthy living
An area of concern is an increase in health-related issues. For example, we may see a rise in sedentary lifestyles as people spend more time at screens, or even possibly leaving the real world for more fulfilling virtual ones. A sedentary lifestyle can lead to musculoskeletal diseases, myopia, obesity, and many more health issues. This is a serious concern for individuals’ health and also would create a large burden on healthcare systems. One solution could be some sort of compulsory physical activity program that is supported by AI, some minimal activity to keep people’s bodies healthy. With the use of general AI, this could become individualized training routines, along with meal plans to help people keep in good shape.
7. Sustainable housing and transport
AI will enable us to understand the impact of products and services on the environment including climate change. Today we use Life Cycle Assessment models to understand the impacts of our consumption in a holistic way, from the extraction of raw materials and the production of products or services, via their use by humans, to waste management. However, there are many unknowns and approximations. AI can help analyze and understand the impacts much more accurately and at a systemic level. This would allow us to optimize the products and services in order to minimize the negative impact on our environment.
AI can help us significantly improve the environmental efficiency of everything we do, starting with how we consume and produce. This is particularly important because so far, our efforts to improve our well-being have brought about negative externalities in terms of the damage to our planet, to our own ecosystem, and in some cases, to our own well-being (think about air pollution and human health). AI can help us overcome our myopia as well as better understand the behavior of dynamic, complex, and uncertain systems – from particle level to species interactions – and thereby arrive at better ex-ante assessments of our impact on the environment and increase the probability of bringing about desired system-level changes – capacities that are simply beyond our human ability.
AI is already able to help reduce waste by enabling autonomous sorting of household and construction waste into piles of valuable materials. AI also already assists in designing new materials and products and even new molecules. AI will be deployed to accelerate these processes further and assist humans in identifying novel ways of meeting needs – often possibly redesigning related supply chains – so they are ecologically beneficial and enhance human well-being. For example, the significant reduction of energy consumption of housing or designing transport infrastructure in a way that allows the transport of goods or food waste. AI will help us in dealing with water scarcity, solving species protection and maintaining biodiversity.
8. A new economy
The structure of the labor force will change beyond imagination. Many currently existing jobs will be automated, however, AI will also create new types of jobs (we covered this topic in greater detail in this blog spot). Although we cannot predict exactly what the new jobs will be, they will be significantly influenced by the changing societal values discussed above. With material abundance, jobs could be linked to activities that increase social good. For example, specific new jobs could be created to converse with the elderly. People could be rewarded for activities that benefit their communities and that benefit their own personal wellbeing. Both societal improvements and also self-improvement will be appreciated and supported.
One of the most talked-about is the reduction of humans as workers as we move towards a more efficient economy. Yuval Noah Harari even talks about making many people irrelevant.
To some extent, the coronavirus may have sped up these processes as well. The automation will probably accelerate as it can help deal with the global pandemic by ensuring the provision of crucial goods and services with few employees, thus reducing exposure to the potentially deadly virus. Also, while the idea of a universal basic income, or other forms of economic support such as conditional income linked to re-training as discussed above or in return for voluntary public or community service, may have been seen as very far away, currently millions of people across the globe start to be out of work and governments are quickly preparing support packages. AI-enabled automation will allow the restructuring of the tax base so the government will be able to finance these new structures.
With the rise of AI, we will see numbers losing jobs to automation, but it is important that we have these structures in place to support people laid off and help them into new and fulfilling roles.
9. Helping developing countries
As AI gets stronger, so does the inequality between the so-called developing countries and developed countries. As Kai-Fu Lee emphasizes in his book, a large advantage for developing countries is their cheap labor and with the rise of automation, this advantage could be taken away as companies and governments move their production back home to their fully-automated manufacturing facilities. Therefore, it is important that the developing world is allowed to benefit from AI technology in the same way as the developed world is. The advances mentioned could give hospitals and schools in developing countries access to top-quality healthcare and education. This could speed up their development allowing them to catch up with more developed countries. This will include catching up also in terms of individual education and capabilities.
AI could also help deal with migration pressures. According to the Economist when migrants move from a poor country to a rich one, they typically make three to six times as much money as before. According to one estimate if everyone who wanted to migrate were allowed to do so, the world would be twice as rich. AI will enable “virtual” migration allowing individuals to realize their potential and truly cooperate globally.
With the advent of general AI, it is likely we will see both negative and positive social and economic impacts. However, it is up to the creators of AI, as well as policymakers, social scientists, philosophers, civil society and other private and public stakeholders to start shaping the future for a more inclusive world that can use the power of technology to benefit people and communities rather take advantage of them or make them even irrelevant.