May 2018

Who Will Win the AI Race? Part II: The European Way

In a previous blog, I compared the approaches from China and the US as they compete in the global AI race. In short, China proposed a government-led approach while the US is leaning on a business-led approach. The European approach represents an attempt to balance both business’ and government’s efforts in directing AI innovation, therefore showing a third way to compete in the global AI race.

Great Britain recently announced a national AI strategy. In a mixture of private, academic and government resources, the country is pledging $ 1.4 Billion in investment. A remarkable piece of the plan is allocated funding for a Center for Data Ethics. The center will develop codes for safety and ethical use of machine learning. Another noteworthy part of the plan is the initiative to fund 1,000 new PhDs and 8,000 teachers for UK secondary schools. This move will not only spur further innovation but also ensure the British workforce is prepared to absorb the changes brought by AI developments. It is essential that governments plan ahead to prepare the next generation for the challenges of opportunities of emerging technologies like AI. In this area, the UK’s plan sets a good precedent for others countries to follow as they look for ways to prepare their workforce for future AI disruptions. Such moral leadership will be a guide not only to European institutions but also help companies worldwide make better choices with their AI technologies. This perspective is essential to ensure AI development does not descend into an uncontrolled arms race.

In the European Union, France has also announced a national plan following a similar approach as the UK. Beyond the mix of private and government investment to the total of 1.5 billion euros, the country is also setting up an open data approach that both helps businesses and customers. On the one hand business can look at a centralized place for data on the other customers get centralized transparency of how their data is being collected and used. If executed well, this central data place can both provide quality data for AI models while still ensuring privacy concerns are mitigated. The strategy also includes innovative ideas such as harnessing the power of AI to solve environmental challenges and a narrow focus on industries that country can compete in. Similar to the British approach, the French plan also includes funding for an Ethics center.

While Germany has not announced a comprehensive plan to date, the country already leads in AI within the automotive industry. Berlin is considered the 4th largest hub for AI startups. An area in Southern Germany known as Cyber Valley is becoming a hub for collaboration between academia and industry for AI. Even without a stated national strategy, the country is well-positioned to become a hub of AI innovation for years to come.

These countries individual strategies are further bolstered by a regional strategy that aims to foster collaboration between countries. Earlier this year, the European commission pledged 20 Billion Euros over the next 2 years for the 25 country bloc. It proposed a three-pronged approach: 1) increase investment in AI; 2) prepare for socio-economic changes; 3)Devise an appropriate ethical and legal framework. This holistic approach may not win the race but will certainly keep Europe as the moral leader in the field.

Conclusion

This short survey from these two blogs gives us a glimpse of the unfolding global AI race. The list here is not complete but represent three different types of approaches. In an axis of government involvement, China is at one extreme (most) compared to the US on the other (least). Europeans countries sit somewhere in the middle. In all cases, advances in AI will come from education, government and private enterprise. Yet a nation’s ability to coordinate, focus and control the development of AI can be the difference between harnessing the upcoming technological revolution for prosperity of their people and those that will struggle to survive its disruptions. Unlike previous races, this is not just about military supremacy. It touches every aspect of society and could become the dividing line between thriving and struggling nations.

Furthermore, how countries pursue this race can also have global impacts on the application of AI. This is where I believe the European model holds the most promise. The plans put forth by France and the UK could not only ensure these countries geo-political position but could have benefits for all nations. The regional approach and focus can also yield significant fruits for the future. Tying AI development efforts with ethical principles and sound policy is the best way to ensure that AI will be used towards human flourishing. I hope other countries follow their lead and start anticipating how they want AI to be used inside their borders. The true winner of the global AI race should not be any nation or region but humanity as a whole. Here is where France’s intention to use AI innovation to address environmental challenges is most welcome. When humanity wins, all countries benefit and the planet is better for it.

Who Will Win The Global AI Race? Part I: China vs USA

While the latest outrageous tweet by Kanye West fills up the news, a major development goes unnoticed: the global AI race for supremacy. Currently, many national governments are drafting plans for boosting AI research and development within their borders. The opportunities are vast and the payoff fairly significant. From a military perspective alone, AI supremacy can be the deciding factor for which country will be the next super-power. Further more, an economy driven by a thriving AI industry can spur innovation in multiple industries while also boosting economic growth. On the flip-side, a lack of planning on this area could lead to increasing social unrest as automation destroys existing jobs and workers find themselves excluded from AI-created wealth. There is just too much at stake to ignore. In this two-part blog, I’ll look at how the top players in the AI race are planning to harness technology to their advantage while also mitigating its potential dangers.

China’s Moonshot Effort for Dominance

China has bold plans for AI development in the next years. The country aims to be the AI undisputed leader by 2030. They hold a distinctive advantage in the ability to collect data from its vast population yet they are still behind Western countries in algorithms and research. China does not have the privacy requirements that are standard in the West and that allows them almost unfettered access to data. If data is the raw material for AI, then China is rich in supply. However, China is a late-comer to the race and therefore lacks the accumulated knowledge held by leading nations. The US, for example, started tinkering with AI technology as early as the 1950’s. While the gap is not insurmountable, it will take a herculean effort to match and then overtake the current leadership held by Western countries.

Is China up to the challenge? Judging by its current plan, the country has a shot. The ambitious strategy both acknowledges the areas where China needs to improve as well as outlines a plan to address them. At the center of it is the plan to develop a complete ecosystem of research, development and commercialization connecting government, academia and businesses. Within that, it includes plans to use AI for making the country’s infrastructure “smarter” and safer. Furthermore, it anticipates heavy AI involvement in key industries like manufacturing, healthcare, agriculture and national defense. The last one clearly brings concern for neighboring countries that fear a rapid change in the Asian balance of power. Japan and South Korea will be following these developments closely.

It seeks to accomplish these goals through a partnership between government and large corporations. In this case, the government has greater ability to control both the data and the process in which these technologies develop. This may or may not play to China’s advantage. Only time will tell. Of all plans, they have the longest range and assuming the Communist party remains in power, the advantage of continuity often missing from liberal democracies.

While portions of China’s strategy are concerning, the world has much to learn from the country’s moonshot effort in this area. Clearly, the Chinese government has realized the importance and the potential for the future of humanity. They are now ensuring that this technology leads to a prosperous Chinese future. Developing countries will do well to learn from the Chinese example or see themselves once again politically subjugated by the nations that master these capabilities first. Unlike China, most of these nations do not count on a vast population and favored geo-political position. The key for them will be to find niche areas where they can excel to focus their efforts.

US’ Decentralized Entrepreneurship

Uncle Sam sits in a paradoxical position in this race. While the undisputed leader, having an advantage in patents and having an established ecosystem for research development, the country lacks a clear plan from the government. This was not always the case. In 2016, the Obama administration was one of the first to spell out principles to ensure public investment in the technology. The plan recognized that the private sector would lead innovation, yet it aimed at establishing a role for the government to steward the development and application of AI. With the election of Donald Trump in 2016, this plan is now uncertain. No decision has been announced on the matter so it is difficult to say what role the government will play in the future development of AI in the United States. While the current administration has kept investment levels untouched, there is no resolution on a future direction.

Given that many breakthroughs are happening in large American corporations like Google, Facebook and Microsoft – the US will undoubtedly play a role in the development of AI for years to come. However, a lack of government involvement could mean a lopsided focus on commercial applications. The danger in such path is that common good applications that do not yield a profit will be replaced by those that do. For example, the US could become the country that has the most advanced gadgets while the majority of its population do not have access to AI-enabled healthcare solutions.

Another downside for a corporate-focused AI strategy is that these large conglomerates are becoming less and less tied to their nation of origin. Their headquarters may still be in the US, but a lot of the work and even research is now starting to be done in other countries. Even for the offices in the country, the workforce is often-times foreign-born. We can discuss the merits and downsides of this development but for a president that was elected to put “America first” his administration’s disinterest in AI is quite ironic. This is even more pressing as other nations put together their strategies for harnessing the benefits and stewarding the dangers of AI. For a president that loves to tweet, his silence on this matter is rather disturbing.

The Bottom Line

China and US are currently pursuing very different paths in the AI race. Without a clear direction from the government, the US is relying on private enterprise’s to lead the progress in this field. Given the US’ current lead, such strategy can work, at least in the short-run. China is coming from the opposite side where the government is leading the effort to coordinate and optimize the nation’s efforts for AI development. China’s wealth of centralized data also gives a competitive advantage, one that it must leverage in order to make up for being a late comer.

Will this be a battle between entrepreneurship and central planning? Both approaches have its advantages. The first counts on the ingenuity of companies to lead innovation. The business competitive advantage for AI leaders has huge upsides in profit and prestige. It is this entrepreneurial culture that has driven the US to lead the world in technology and research. Hence, such de-centralized effort can still yield great results. On the flip-side, a centralized effort, while possibly stifling innovation, has the advantage of focusing efforts across companies and industries. Given AI potential to transform numerous industries, this approach can succeed and yield tremendous return.

What is missing from both strategies is a holistic view of how AI will impact society. While there are institutions in the US that are working on this issue, the lack of coordination with other sectors can undermine even the best efforts. In this range of centralized planning and de-centralized entrepreneurship, there must be a middle ground. This is the topic of the next blog, where I’ll talk about Europe’s AI strategy.

Travelers Theology: Wrestling With a Powerful AI God

Recently, I was browsing through new shows on Netlflix when I stumbled upon Travelers. The premise seemed interesting enough to to make me want to check it out. From the very first episode, I was hooked. Soon after, my wife watched the first episode and it became a family affair. Starring Erick McCormack (Will & Grace) and directed by Brad Wright (Stargate), Travelers is a show about people from a distant future that come to the 21st century in an effort to change history to re-write their present.

You may wonder “Nothing new here, many shows and movies have explored this premise.” That is true. What makes Travelers unique is how they arrive in the present and how they explore emerging technologies in a thoughtful and plausible way. They travel back to time by sending the consciousness of people from the future into the bodies of those who are about to die in the 21st century. Having the benefit of knowing history allows them to pintpoint the exact time for arrival which makes for some pretty interesting situations (a wife about to be killed by her husband’s abuse, a mentally-challenged woman about to be attacked by robbers and a heroin-addict about to be overdose). The travelers then continue the life of their “host” making those around belief that they are still the same person that died.

Spoiler alert – the next paragraphs will openly discuss plots from the show

By the end of first season, we learn about the pivotal role AI plays in the plot. Throughout the first episodes, the travelers keep talking about “the director” who has a “grand plan.” That becomes their explanation for carrying out missions when they cannot understand why they are doing what they are told. They also follow 6 rules to ensure their behavior limits their interference in the past. At first, the viewers think they are talking about a person who is leading the effort. In the last episode of season 1, we learn that “the Director” is actually a Super Computer (a Quantum Frame) who is able to consider millions of possible scenarios and therefore direct travelers to their assign missions. We are really dealing with AI God, who is quasi-omniscient and demands human’s trust and devotion.

Exploring Rich Religious Imagery

While the show explores religious imagery throughout, this aspect comes to the forefront in episode 8 of Season two. In it, one of the travelers (aptly and ironically named Grace), is to be judged by three judges (programmers). The setting for that: a church. As they gathered in the sanctuary, the “Trinity” of programmers initiates proceedings under the watchful eye of the Director (through a tiny camera that records the event). Grace, an obnoxious traveler who is devoid of social skills, is charged with the crime of treason for taking action on her own initiative in direct challenge to the grand plan.

As the judgement unfolds, scenes that juxtapose the programmer judges with an empty cross in the background reinforce the explicit religious connection the writers are making here. Throughout the hearings, Grace insists that her actions, even if unorthodox, were only to save the Director. Yet, she is surprised to learn that the Director itself had summoned her judgment. She seems disappointed at that, wondering how would the Director judge her if it knew her intentions. This is an interesting assertion because it implies that the director actually knew her thoughts, raising it to the level of a god.

Grace is found guilty by the programmer trinity and is handed over to the director for her sentencing. They speculate that she will be overwritten. That is the worse punishment, which means she would not only die in the 21st century but her consciousness would cease to exist. It is the theological equivalent to eternal death or annihilation.

The next scene is probably one of the most profound and provocative of the whole show so far. Grace goes to a small room where she faces three large screens from where the Director will speak directly to her. This is the first time in the show where the audience gets to see the Director in action by itself rather than through messengers.

While she is no longer in the sanctuary, the room still has an empty cross in the background and evokes the idea of a confessional booth. At that point, I was really curious to know how they would portray the director. What kind of images would she see? Would it be of the machine itself or something else?

No machine but human faces show up in the screen. They are all older and seem to be in some type of life support. At times, they seem to represent Grace’s parents but that was not clear. In this climatic scene, Grace finds forgiveness from the Director and is not overwritten. The machine communicate divine qualities through human faces. Grace finds peace and absolution and re-affirms her trust and devotion to the Director. In short, she experiences a theophany: a watershed personal moment that reveals a new facet of the divine being to a human receiver.

Photo by Bruno van der Kraan on Unsplash

A New Perspective on Omniscience

What to make of this? I must say that when I first learned of Levandowski’s efforts to create an AI religion, I discounted as sensational journalism. Surely there is a fringe of techno-enthusiasts that would follow that path. Yet, I could no see how such idea could be appealing to a wider audience. Seeing Traveler’s religious treatment of AI have made me re-think. Maybe an AI religion is not as far-fetched as I originally thought. An advanced AI bolstered by powerful hardware and connected to a vast digital history of information could indeed do a great job in optimizing timelines. That is, it could consider a vast amount of scenarios in ways that are unfathomable to the human mind. This could make it quasi-omniscient in a way that could elicit a god-like trust from humans. One could say such arrangement would be the triumph of secular science replacing a mythical god with a technological one.

From a Judeo-Christian perspective, an AI god would be the epitome of human idolatry. People worshiping idols except that for calf images are replaced by silicon superstructures that actually can hear, speak and think faster than any human. This would be an example of idols in steroids. As a firm believer in the benefits of AI, I do worry about human inclination towards idolizing tools. As a Christian, I owe my allegiance to a transcendent God. AI can only be formidable tool but nothing more.

Yet, the prospect of an AI god is still interesting in that it may helps us understand a transcendent God better. How so, you may ask? Religion is often defined by powerful metaphors. For some monotheistic faiths, God is a father. Such metaphor has obvious benefits as it elicits image of authority, provision and comfort. I wonder if using a powerful AI as a metaphor could reveal part of divinity that we have not explored before.

In a previous blog, I suggested that AI offered a paradigm of partnership for religion as opposed to blind obedience. Reflecting on Travelers’ portrayal of an AI God sheds light into the aspect of God’s omniscience and wisdom. A timeless being with infinite “processing capacity” could very well consider all the possible alternatives and come up with the best one that leads to the best outcome (to whatever that best is defined). In computer science terms, the best is defined by an objective function – basically the goal you are trying to achieve.

How is that different from previous views of omniscience and wisdom? In the past, omniscience was seen as the idea that God knows what decision we will make and therefor ultimately knows the future. In some traditions, this idea was amplified into the concept of Predestination. The problem with such approach is that it limits God to one outcome and makes humans “automatons.” In other words, there is really no choice or risk – everything is pre-determined from the beginning. I suspect this view of God was heavily based on our own human mind that cannot consider more than 1 scenario for the future at a time.

What if God’s omniscience was more like the Super AI knowledge that is able to simultaneously consider multiple outcomes and then guide towards the better one or correct it when that path is undermined? Wouldn’t that be a fuller view of omniscience? This scenario allows for human choice while still attributing superior knowledge and control to God. Furthermore, this metaphor reveals a “smarter” God that is not bound by the one-track linear thinking of humans. Humanity realizes that their choices matter and can create alternative futures. Even so, they still have the comfort of a God who can see through all this, and guide it from a perspective that can consider manifold outcomes.

Such God would certainly be worthy of human obedience, awe and praise.

Month: May 2018