Would AI be better at governing than politicians? — Branka Panic, AI for Peace
Interview with Branka Panic, Founder and Executive Director of AI for Peace with Ciprian Stănescu, Founder and Editor of Future Magazine
First of all, what is AI? Everyone is talking about it, but few actually understand it.
There is no single agreed and accepted definition of AI. And yet, it is widely accepted that AI is all around us and shaping our lives substantially. From Google search, maps and Siri, to IBM’s Watson and Project Debater. AI is typically defined as the ability of a machine to perform cognitive functions we associate with human minds, such as perceiving, reasoning, learning, and problem solving. To be able to define and understand what AI is, we first need to understand the distinction between different levels of AI. Artificial Narrow or Weak AI performs only single narrow tasks, like facial recognition, internet searches or driving a car. On the other side, Artificial General (AGI) or Strong AI is closer to human intelligence, developed on the human brain as the model. AGI will be able to, in theory, successfully perform any intellectual task equally to human being. As such, it is the long-term goal of many researchers, although they don’t agree how likely it is, we’ll have AGI soon. There is also a third type of AI, called superintelligence, more powerful and sophisticated than a human’s intelligence, capable of achieving consciousness and surpassing humans at absolutely everything.
“AI will come and take our jobs.” We hear that a lot in clickbait media - what is your take on this?
There is a lot of disagreement about this topic between pessimistic voices who claim that AI will take our jobs and more optimistic ones, claiming AI will create new jobs and more jobs than it displaces. No matter how much we disagree on this, we can agree about one thing: AI will definitely bring a lot of disruption. I personally believe the optimistic view is the more realistic one. Automation will “take” some jobs away, but it will also create some (or many), and some that are beyond our ability to even imagine at this point. Before the internet, it was hard to imagine that web designers or data scientists will be in demand. Let’s look at the medical field today, where 60% of medical staff expect, with the help of AI, to spend 70% more time performing procedures vs. administrative work. Robots will become part of our everyday life sooner than we think. Being in the midst of the coronavirus crisis we see how hospitals staffed with robots in Wuhan in China protect medics, taking patients' temperature, delivering meals and disinfecting the facilities, reducing doctors’ exposure to the virus, protecting them and freeing their time to be where they are most needed. AI has proven valuable every step of the way in this crisis, from detecting the virus outbreak, revealing patterns of detection, diagnosing the disease to developing treatments and vaccines. Here is another example of AI applied for helping radiologists reading and interpreting possible cancers that cuts reading times in half. This means radiologists will be able to interpret more images in the same amount of time and make diagnoses more efficiently. AI will not replace doctors, but doctors who use AI will replace those who don’t. That’s why it is crucial for all of us to know what AI is, and to come back to your first question, not only know but understand properly, in order to use it with maximum benefit.
If a robot (using AI) can do repetitive tasks or imaginative tasks better than a human, what else can a human do?
A lot! Empathy, critical, innovative and imaginative thinking, emotional intelligence, the ability to recognize the problem and figure out a solution are all important traits of people that (still) set them apart from technology. However, AI is becoming increasingly better in many of those traits as the number of companies working on improving those traits are increasing. Let’s take the example of empathy, the one I truly believe people will be better at expressing for a long time. However, with the growing development of chat-bots, injecting emotional intelligence into chatbot-based communications has become a development priority. It’s also interesting that, with the growing interest and requests for ethical AI, “right to explanation” of algorithmic decisions is becoming one of the priority requests. That means that although the decisions are based on the algorithms, the “black box” type of decision making, technology that delivers answers while hiding the process that leads to them, is increasingly not acceptable anymore. This is where humans have to take over and be able to “unpack” the black box. Humans are the ones who not only explain, but also recognize mistakes, diagnose errors and help avoid similar mistakes and unintended consequences in the future. Researchers at Stanford are introducing the human-in-the-loop approach that reframes an automation problem as a Human-Computer Interaction design problem. This system would lead to rephrasing your initial question, instead of “what else can humans do if intelligent machines create automation?” but rather “how do we go beyond simple automation and use AI to augment human capabilities, incorporating human interaction and judgment, in order to design more meaningful AI systems?”.
Would AI be better at governing than politicians? What would be the principles behind such an AI?
Politics is often considered as one of the world’s least trustworthy professions. Learning from experience and previous examples, we often question politicians’ motives, honesty and sincerity. So, if we can remove all the flawed traits of human beings from political decision making, should we reject this option? In fact, a survey conducted last year in eight European countries showed that 25% of Europeans want AI to replace politicians. Imagine a world where the AI government is making rational, fair and evidence-based decisions. This government would also be faster, with superior analytical capabilities, and immune to corruption and unethical behavior. The truth is, that at the current stage of development, AI is not able to do so. But sometime in the future, most certainly yes. AI is currently only as good as the data it uses, and data often reflects our biases. Therefore, at this stage AI will not always reduce human flaws, but sometimes even amplify them. But this is not a reason to disregard this possibility in the future, while working in correcting our and machine bias. While some are imagining and working towards replacing politicians with AI, others are transforming the entire political system through the power of AI. This idea is suggesting bypassing politicians by empowering citizens to create personalized AI representatives that participate directly in democratic decision making. Instead of having a representative that represents you and millions of other people you can have a representative that represents only you. In this future parliament has as many parliamentarians as we have citizens, and those parliamentarians are able to read every law and vote on each one of them.
Let us go far into the future. 2222: how would a world with a superior AI look like in terms of these 3Es: ethics, efficiency and entrepreneurship?
Welcome to 2222, the age of superintelligence. Most of the jobs are performed by machines, while people, through a universal basic income, are pursuing their lives unburden by the need to earn a living. Universal medical insurance is free for everyone across the world. Through gene editing and bio engineering we live long and healthy lives, free from illness, pain and aging. Poverty, hunger and disease are part of history. Nanobots are connecting our brains to the cloud, leading to brain-to-brain communication and instant learning. And all that thanks to the work of AI. At this point we are co-existing with intellect that is much smarter than the best of us, literally in every field, from science to social skills. Goals and values of such AI are aligned with ours. Ethics now encompasses not only guiding principles for human behavior, but for artificial intelligence as well. The very same principles protect humans and super intelligent AI, as AI is deemed as a person and given the rights that every human enjoys.
This is certainly the optimistic scenario of coexistence of super humans and super intelligence. The reality of 2222 may be close or very far away from this scenario. The way to the future is probably paved with many more AI winters to come, and probably many more human-made disasters and crises, that will slow scientific and development progress. Super intelligence may happen in decades, centuries or never. AI experts disagree and we simply don’t know. But most certainly the future will be heavily influenced by decisions we as humans make on this path. We get to choose what AI we build. If there is a future which is catastrophic for the human race, we should steer away from it. The sooner we start thinking about solving safety and security problems of superintelligence of the future, the more likely it is that this future will not be a catastrophic, but a utopian one.
Branka Panic is the Founder and Executive Director of AI for Peace, a nonprofit ensuring artificial intelligence benefits peace, security and sustainable development and where diverse voices influence creation of AI and related technologies. Branka is a passionate advocate for positive peace, with 10 years of experience in policymaking, working with governments and think-tanks across the globe. She is a co-founder and Board Member of the Center for Exponential Technologies, connecting policy and the tech world. She is author of “AI EXPLAINED Non-Technical Guide for Policymakers”.
Twitter: @Branka2Panic; Medium: @bpanic; LinkedIn: BrankaPanic