IGF 2017 - Day 1 - Room XII - WS12 Social Responsibilities and Ethics in Artificial Intelligence


The following are the outputs of the real-time captioning taken during the Twelfth Annual Meeting of the Internet Governance Forum (IGF) in Geneva, Switzerland, from 17 to 21 December 2017. Although it is largely accurate, in some cases it may be incomplete or inaccurate due to inaudible passages or transcription errors. It is posted as an aid to understanding the proceedings at the event, but should not be treated as an authoritative record. 



First is a computing power, right?  So the investment of oh, the parallel and disputed computing, technology really enables us to ‑‑ the process power can get scaled to up three sizes these days.  The size of the center just grows proportionately to the size of the application.  Second, it's a data.  The availability of big datasets derived from the internet error that those data actually enclose tremendous information about human behavior's knowledge. 

And now third is ‑‑ that especially in supervised learning that has enabled us to leverage precisely the computing power to learn from the data sets, that tremendous information about human behaviors and information.  Also in addition today right so the ubiquitous internet and the social networks, mobile networks, really has served as a very efficient infrastructure that for quick dissemination of air technologies were never some good products out there.  We can quickly reach and be embraced pie huge mass of people.  So the impact basically can be very quick and very large.  So just use U.S. and China as example.  I probably will see lots of most of the sensation that happens in these two countries recently. 

But AI really has become the fastest growing sector in the industry by almost all measures, best and size and starting companies and so on, business and products and so on.  So but both country actually also if you look at it more closely, they have their own strengths and so on.  And they're also very different in nature.  So for technological standpoint, we can see the United States is very strong in the traditional fundamental technology for AI, such as chip design and manufacturing, server operating systems, therefore big data center, and also computing platforms such as for deep learning and force.  Whereas terrorize very much focuses on applications and products of AI such as speech recognition, face recognition, ‑‑ very much probably due to its large scale market size and marketing there's lots of. 

In terms of from the research of AI, the both the United States and China lead the world.  The rest of the world by a large margin in terms of papers published or patents filed each year.  However, the United States university is a company focused more on the fundamental theory, machinery theory, technologies, and also systems level.  Well, AI researching chart again is the most of the application treatment about as many of the applications of mission learning and other source.  So in terms of investing AI, based on the numbers of ‑‑ venture capital is in the United States is certainly larger in China, but China is really catch up very quickly. 

Already in the same magnitude as America.  And again the investment in U.S. covers almost all areas of AI technology, from the hard wears all the way to the hardware system, softwares, and all the way to the applications and problems.  Whereas China shows strong focus on applications and end products, especially on consumers level, for mobile networks.  Nevertheless, the Chinese company has made international strategy to strengthen its position on artificial intelligence and a plan to pouring tremendous government support, not just in financially but also a policy and so on in the next decade. 

So we probably will see a potential change of the balance in the U.S. and China.  So ‑‑ and also so ‑‑ so we will know that the process of technological amassment is actually something that can take a long time.  And it is such a transformation also that acceptance of technology by users and scale requires really a gathering of all the stakeholders, all UN nations.  ‑‑ and its applications and how the technologies can potentially transform better our lives. 

In the past 20 years since I graduated from UC Berkeley, I have witnessed the development of technology.  This technology were invented 30 years ago but was thought really ready for the general applications probably until now.  Today we really see that more and more companies world wide that applying at technologies and a scale to mass market.  For example, speech recognition, face recognition, striving, are really ‑‑ those speech or facial recognition already integrating our cell phones, smart phones, and also that smart mobile payment through China and many other methods has become part of our daily life.  This is just beginning of ‑‑ in fact, we really just see this as the beginning of the era. 

We work on the witness very quick de‑employment of many many AI technologies at massive scale and years to come.  So today I'm very, you know, like to think that IGF and also university of Zurich, other chart of academy ICT, for creating this opportunity to gather people here to discuss the AI, not only its impact on the IT industry, but also its far reaching implication on our society and the world. 

We hope that just like the internet, through really a multinational and multi‑stakeholder discussions and collaborations we will be able to reach international consensus for the potential governors of AI technologies so that it can really best benefit us and our next generation of the future.  That's all, thanks. 


Despite architecture of this room and the panel itself, which doesn't invite for a dialogue really or a lot of interaction but is more designed for statement after statement, we will not do the statement after statement.  We will really try to have a conversation and we have a fantastic line‑up of speakers these customs which were already introduced by Chia.  But I thought maybe we could start with a quick round of self introductions.  And if I may ask my colleagues just maybe briefly to highlight the few things that you're working on, how they connect with the panel here today, and the topic of social responsibility and defects in AI.  And secondly, maybe after the very helpful high helve introduction to share with us what's your favorite AI based app or technology that you are using and why do you like it.  So Danit, maybe start with you please.

My favorite app is I think is the Google translate.  Which is the most ‑‑ which I think is the only one that we can use in China.  But actually I use Google translate a lot, and sometimes we just scan it and the picture will automatically comes to the language I want.  But I think it still has some problems.  When we scan a lot of sentences or paragraphs.  So I think Google translate maybe also has some rooms to improve.

My name is Irakli.  I'm working for the United Nations, namely for eunuch in the heading they just created center for artificial intelligence of the robotics.  And just a very few words about the center.  The center is focused on actually bringing multi‑stakeholder corporation into fruition, so bringing together the governments, private sector and academia and other interest in it.  Also running educational training mentoring session programs for different stakeholders, including government officials, diplomats and others. 

And at the same time soon we're going to start doing the country assessments and matching the available technology to solve the world's pressing problems like the mis‑sustainable development goals and so on and so forth.  So as far as the favorite app, yes I have a lot of favorite apps.  Of course any translation app really is helpful or the hotel suggestion, I'm also travelling a lot and having stayed in different hotels and having good consumption.  My daughter's favorite app is musically, they make a little music clip videos and that app just happened recently.  It's Edward by Vyvanse and my daughter was very happy when I told her I know people very well who own this app.

We heard a lot from our opening remarks about the very quick and large impact potential of AI, and we really based on the title of the session, look at the social impact and the social responsibility associated with that.  The impact that we'll have on jobs, the way we live, the way we work.  Put I think in part of the major focus of our noble initiative is we can't lose site of that human impact and wellbeing.  Hopefully we can talk a little more about that.  Thank you.

Their probably favorite app is to play with Siri, the speech recognition.  So they ask all kinds of questions, you know, hike what's a score of sea hawks, they lost the game or not or what's the time and how to get movie tickets online.  It's just amazing to see how they have fun with the cell phone, right.  And just total hits me that the technology going to change your iPhone.  They're going to have a completely different life than ours.  Their knowledge, their social behaviors, and including their customs will be very different from ours.  We have to think ahead about the impact, not just us, on our future generations.

What are for you maybe starting with some of the biggest opportunities that you see in terms of social impact of the next life of technologies and particularly also since you have to benefit to live both in China as well as a you have deep understanding of what's happening here in Europe and U.S.  how does that play out globally, some of these opportunities?  What are your reflections?

But basically one of the ‑‑ or maybe two sort of bigger issues I would really like to see in making breakthroughs would be with the health and applications in the healthcare and making really breakthroughs there and also in educating poverty.  And I think in both cases, if we manage to do that, that would pee a really great achievement in the short‑term, but in the long‑term, I guess all of the goals would definitely benefit from any of the breakthroughs basically from that tool, what AI is to be applied to that and having benefits out of it.

The document says it's trying to seize a mage major strategic opportunity to advances development of AI, and this new plan, which will be implemented by a new plan promotion office within the ministry of science and technology, out lies Chinese ‑‑ China's objective for advancing AI in thee stages, which first by 2020, China's overall progress of applications and Chinese should keep pace with advanced world level while AI industry becomes important economic growth point. 

And the second step is by 2025, China should have achieved major breakthroughs in AI to reach a leading level with AI becoming a primary driver for China's industry advances and economic transformation.  Ultimately, by 2030, China intends to have become the world's premier AI invasion center.  At that point, China believes it can achieve major breakthroughs in research and development to occupy the commanding heights of AI technology.  So if we look at this plan, I think it's not just a technical or industrial development plan but also includes social construction, institutional reconstructing, global governance, and other aspects.  In other words, we are facing ‑‑ the task we are facing is not to achieve revolutionary technological breakthroughs in a particular field or industry, but to vigorously promote comprehensive changes resulting from technology codevelopments.  So I think in China, the major challenge to Chinese people is to achieve the economic reconstruction.  So AI and also together and also the digital revolution has becoming a major economic growth point to realize the economic reconstruction and also to help people make a with a better life.  Thank you.

And I was wondering, would you be willing to share a little bit more, yes, there are institutions, yes, there are systems.  But what does it do to us as human beings.  And again staying a little in the opportunity, asking a little segue to the challenges of course later on.  Please.

But it has a tremendous opportunity, when you think about agriculture and health and learning and education.  But also we need look at it from the trust factor.  There's concerns about what's happening to the data because of the massive amount of data capture and use, which could be extremely impactful and very beneficial.  But people aren't necessarily informed about how data's being used or it's not being used in a good way, in a sense that it can really expedite developments and the benefit of AI.  We can run into some challenges there.  And when we look at the impact of AI with the evolution of jobs and thing of that nature, we need to also be very cautious of that also being in the impact on human, when it comes to that. 

We are going to be in the transition period.  There's going to be an evolution where there might be some gaps as people in the job market and education may not be necessarily sealed or have the capacity or there's a transition where AI might be impacting some of the jobs, and there's going to be job transition.  And that can really impact well being and how people, you know, not only from the impact on their economic livelihood, but that impacts you emotionally as well. 

So when we're looking at autonomous systems, intelligence systems, looking at from the tremendous opportunity is at our feet.  And what that can mean to our future generations.  But also thinking about the impact on well being, not only from a medical health perspective but your mental well being as well.

So I think there's tremendous opportunity in that.  Today, we hear so much about the channels that are out there with health and mental well being and it's almost like at a cries through many parts of the world.  And I think artificial intelligence, autonomous systems can really help with that situation.  And we sort of build trust into that and when we're building the systems, we're taking into consideration that level of impact.  We really to look at these things very holistically in that sense.

I think that we are kind of maybe stuck in a zero sum game mindset when we think about two countries, winner take all.  And I think my best solution for feasible and sustainable inclusion is to move away from that kind of zero sum game mentality and really make sure that whoever gets the best technology, that's great.  But we really need to start thinking about who is not having the technology and who should be having it and how they could employ it.  Even if you have the best technology, if you don't share it, if you don't profit from it, if you don't distribute it, it doesn't mean anything.

I'm not asking that.  But one of the solutions which we saw could be practical and certainly could contribute to that process and something from, our center we are really going to start implementing is to start the pilot projects in the countries on a practical application of these tools to solve practical problems for the benefit of the countries and the people.  And see that this works and then make sure that the other parts of the world countries will see that these product can be sold like this and scale it up, get others excited, and actually make that move. 

But when the ‑‑ the companies who have the technology available would be actually willing or inclined to share that with the countries which would need to solve particular problems.  So that type of scalable approach, which we thought that could be one of the things which would we would definitely like to try out in practice.

And that there would need to some investment, not just in terms of technology, but also in a human talent.  I think China has done ‑‑ the is a lot of really accident that China was able to catch the a little bit, you know, 93 very closely with the Ai technology because they're not highest being investing a lot in the computer industry for a long time, starting for the internet H has been trying.  China wasn't behind but really trying to catch up with a lot of the companies investing in the past 10, 15 years.  And when suddenly some of the technology becomes applicable, this company will have the talent, people and resource to put those technology in mace and scale it to mass market. 

So that might be a lesson can be probably learned by other nations as well, if they want to probably emulate the process to catch up with the technology. 

And also opportunity for all the nation is that great things about AI that's mentioned before our discussion is that if you do not invest, you are definitely left behind.  But if you do invest, we able to really need other people by a large margin.  And also because of that, you see there's a very interesting ways to patent AI technology, all the major companies hike to share.  So there's a lot of source.  So that's an article point of view, right?  I think also.  So all the platform, especially computing platforms, and a lot of them are open source.  So that really allows AI to be dissimilated very easy to many many countries, the benefit of many many nations if those nation are willing to learn, willing to.  So the bar is actually not as my as any other technologies.  So it's all very prohibitive as other AI technologies, like manufacturing or other.  So that's actually I see a really great opportunities for, you know, all the nations.

So what, you know, looking at it more from a scientist perspective now, or an academic perspective, what methods of social impact assessment do you work with in your centers on the academy and what can we learn from the past?  So is AI something that requires an ecosystem perspective like the environmental movement and environmental impact assessment?  What are the frameworks you're working with, maybe from your perspective?

One of the things which we would like to see is how countries would be using the technology for the benefit of solving different problems, right.  And what I mentioned with the sustainable development cause would be really good assessments to see how much of these technologies would be really able to solve it and how quickly this would happen and how quickly we can bring the message to the larger sort of part of the world that this can be done and this can be done really in a responsible way and this can be done in a way that it could really benefit the countries and benefit the people because that's what end of the days the most important aspect of what we are doing, right. 

So therefore, yes, we should really go on and test out different methods and see how we can bring the benefit.  But at the same time taking into account the risks, but I think that it's something which you would like to discuss likely later in the panel, as well.

So this obviously is a great segue to the challenges part of our panel.  And since Karen you already introduced some of the key words and I liked very much your starting point to look at it as a trust challenge.  Yes, we have a data privacy problem.  We have the question of bias.  So there's a long list of specific challenges, but I think it was very helpful to look at it as a trust issue.  And I was wondering whether you can perhaps expand a little bit on this notion of trust and why it matters and also how we can break it down into chunks of problems maybe.

And you know one of the concerns could be is that if there's a fear and concern about data and technology, not understanding it technology, what's happening to my data, there might be a slower up take of that use of that technology that can really benefit human beings, societies, governments.  So along with addressing these issues, we have to look at it from that overarching concept of trust.  And I know there's many definitions of that, and it's conjectural and cultural to a certain degree as well.  But how we can help fill that gap, if you will, or close that gap, where, you know technology is advancing.  We're coming up with great acts and solutions and systems that can really benefit many sectors ‑‑ all of society, in fact. 

But yet there is sort of this I'm not quite sure what's going on.  For sure, there's definitely a population, probably more western, that, you know, we just sort of give into the technology.  You know, there's convenience factors associated with it.  We just plug in and use it.  But we also have to look at it especially in the concepts of IGF, and one is digital inclusion and internet conclusion.  We are bringing internet to to people who are underserved and don't have access, what other issues are we introducing to that as well.  Part of what our responsibility should be is learning and helping from our learnings when we do that so hopefully they don't experience some of the same discomforts or challenges that many of us have. 

But kind of what it appeals down, when you're talking about digital identity, the use of these technologies, what is that really going to mean to wealthy and into the human factor.  IEEE's mission statement is about advancing technology for the benefit of humanity.  And that takes many forms and many aspects and many industries in every day life, but we are really working hard to ensure that we're focusing on that benefit for humanity aspect and the well being aspect of it.

A lot of diagnosis and image based analysis, prognosis.  So I've learned that when we try to you know swing that or disseminate the technology to those disciplines or fields, something there is barriers between.  So I think we try to get help, work with Harvard medical school, and somehow it goes through.  Basically also has to be trust across other fields.  But also, the automobile industry.  The trust, or the driving technology, would it put them out of the business.  So again we've I hope that people have to get out of this mindset of zero sum gain.  So ‑‑ they will take our job.  Rather, it's really that we should build a trust in the technology so that you actually be able to have a win‑win situation and in fact the industry will benefit from all the new technology rather than you know the job gets taken away. 

So how to use the technology combined with traditional industries to create more opportunities, make the industry more efficient, have a higher standard, qualities and so on.  I think that's kind of a challenge we're facing today, to have AI technology to have bigger impact, different industrial sectors as well as aspect of human lives.

Let's look at the developing world, for example, and how this is going to impact the job market there and then what other sort of chain reactions it's going to have.  We're not only talking with the Asias of the jobs here, which I think that jobs are also very much associated with the issue of migration and this is associated with the issue of security and the peace in the world at large.  So we need to actually take a look at the really larger picture. 

If we look at the job issues, what is happening in the developing world?  We have interaction with many governments in the developing world and discussions related to the automation or technological automation is mostly nonexistent there.  One thing that really needs to be done is to proper understanding to the field there, to the countries, and they would really need to realize that how this is going to impact.  The second thing of course what we need to bear in mind is the rate of automation, how quickly it will go.  And this nobody really knows because nobody would expect that bit coin would become $19,000 within matter of days and so a lot of people are regretting not to buy bit coins or people will regret that they bought it.  Basically the rate of expiration is really unknown and very difficult to predict. 

And the second aspect would be how are we really going to match up our preference with the rate of expiration.  So how quickly will we react to that speed of automation?  We really need to be working very closely with all these developments, but certainly one thing that needs to be done and needs to be done right now and very quickly is creating different understanding of these changes, bringing the knowledge to the countries and to why the spectrum of the world and finding many solutions.  Because right now what we have is the two sets of solutions really at the table.  One is universal basic type income type of solutions like taxation and so on and so forth. 

And the second one is training and education, right?  And basically none of them are pullet proof.  It's very difficult to sustain the universal basic income type of solutions and obviously at call, it has something very interesting in it but it's very difficult in principal to apply. 

The same thing on retraining and education.  On education side, you would need to fundamentally change the entire side of thinking behind that.  When we're talking about how our kids are born into the AIs type of implication, we need to to think about what sort of skills they would require, but this requires really a global movement rather than selective movements.  Therefore, these are massive jobs which would need to be done.

Whether there's ‑‑ and I really couldn't speak to it that definitively.  Whether there's a gap between our education systems and how it's working or preparing our young children as we're going through school, how the education system is using this type of technology could be a gap.  I know when the internet was kind of rising up and still some schools, also there was a gap ‑‑ we were teaching with traditional white boards, black boards, if you will, and ten then technology came on board

>>SPEAKER:  There's still challenges with that because I can speak just from my kid's perspective.  You know, the homework, I saw it increasingly become more online.  You know, so if you weren't ‑‑ didn't have access or you didn't have the equipment in your home, it was a challenge, right?  So same thing.  You know, we have to be cautious of it.  That doesn't necessarily happen with these other advances in technology of autonomous systems or artificial intelligence, that we're not leading this gap in how we're expecting to use it but yet when the children go home or the worker goes home, they don't necessarily have access to it. 

So by doing that, we kind of keep this gap ‑‑ an unfortunate gap kind of in place.  So I don't think there's necessarily one magic Answer two to any of this.  I think it's going to be a lot of institutions and systems and kind of working together, if you will.  So there lie as little bit of a challenge, but I think if we take an example of what we're trying to do at the internet governance form and how we've looked at a multi‑stakeholder approach with that, I think if we sort of apply that to this, we can definitely address some of these challenges.

So what I'm saying is that it turns out there was routine to be concerned but we actually were concerned about some of the wrong problems.  Having said that, Urs mentioned a few issues like privacy and bias.  And mostly the panel is focused on jobs, but I think it's important to address what's out there in the ecosystem.  So it's the kind of thing Elon Musk is talking about.  I wonder if you would kind of address some of the common concerns that people have, where it may be relevant or irrelevant and maybe some of the things you're thinking about that many of us may not even be working about yet.

One of the main things that we have that artificial intelligence does not is creativity.  And this is humanity.  We adapt.  We adjust.  We develop.  So in that sense, I think that's something that a lot of technical people are talking about, is really making sure that we design the technology and talk to the people who designed the technology because it is a human outcome.

So some of the skills or technologies can be easily basically hands‑on some people.  Just like any technology, just hike double edged sword.  You can use it for good or bad.  So that could raise some concern, that some of the missionary capability, if it falls into the hands of wrong people.  And also that can be very easily falls into the hands, unlike any other physical materials of usually much easier to monitor.  So those could be technological concerns I would have for some of the technology falls into the peoples.  So that also raise so people can bridge privacy, bridge security.  You know, people can use it to interior with elections and so forth.

But this is creating a huge issue.  And once you add machine learning technology to it, which is going to serve as an amp fire effect, we might actually end up a day where we will have every day billions of this type of codes created.  And it would be very difficult to deal with.  So this is going to create actually quite a lot of pressure on security of quite a lot of pressure on actually entire topic of our society.

And so that was for people in the future.  I think that's the fundamental routine for the moral panic, for people when people get to know artificial intelligence.  And I think another source of the panic is about the ‑‑ is from the drama and the films that we watched before and the machines take control of the human word and the humans is totally lost.  I think that's also ‑‑ that's also a scenery that it's not happen right now, but we will wonder whether they will come true in the future.  So maybe some fields. 

Another thing is some days ago I watched the TV news.  The female robot, Sophia, was granted the first citizenship by Saudi Arabia government.  And during the interview, Sophia said I wanted a family.  ‑‑ Sophia said I wanted a family.  And I think that's the moral panic, the concern that we got from those films and drama.  So when we come to artificial intelligence, especially for the experts or scholars or researchers, we need to take some forward looking approach to this technology.  So we might concern that if we give some emotions to the artificial intelligence, because I'm not scientist.  I'm not sure whether they will happen.  But if they have the emotion or they come into our social life and get ‑‑ interact with the real ‑‑ with humans, maybe some ‑‑ in some sector, we should strictly forbid those artificial intelligence has quickly used.  Because my field is international studies. 

I also thinking that just like was just said, technology has always been a double edged sword, just like internet.  If those terrorists use artificial intelligence, artificial intelligence has been used for military purposes, what will bring to us.  So I think for those consequences we need to really take a forward looking and concern.  And of course for social scientists, we also need to know what artificial intelligence will develop.  That's my response.

And I have three observation.  You can transform it in a question afterwards.  The first thing is I was in Paris in 1989.  No internet.  No social network.  So big data, no IOT, no cloud and so on.  And I do some clearing at home because I am retired.  And I look at the paper, the accommodation of the ‑‑ I see units ‑‑ ICT and education.  It was four pages, 15 recommendation, and you cannot apply today.  But we don't do anything during 28 years.  We were only distract by evolution of technology.  Everything is already in these recommendation, very old recommendation but quite at the ‑‑ quite for now.  You can learn at home, you can learn at distance, you can have books, doesn't exist in 89.  And so okay. 

So I think we should be bit more wise to take content and a thing like that and not only to see the exponential evolution of technology.  That is my first remark. 

By the way, I have here a nice article from the institute of the future, gives for 2020, 2030 and so on.  And you have a map with six trends, actual trends, and ‑‑ competency to acquire for the future.  Nothing to do with actuality ‑‑ in any country.  Nothing.  Complete ‑‑ educational system nowadays are producing handicapped people because we are not in the right ‑‑ we are not in the right vision.  That is my second point. 

The third point is another silly thing, but I have to say that here when you are speaking about artificial intelligence that I define as the opposite of natural ‑‑.  It's the text that we have written with some colleague in ‑‑ and presented last with this in ‑‑ in June and we are now at the CSTD in the election of the general assembly of UN.  The title of this text is very simple.  It's human digital right and responsibilities.  And we thought that if human being is delegating to AI, to network or robots, or whatever it's the end of the specials.  And we have to keep the power of and the responsibility of using AI, ICT, and thing like that.  Okay.  It's not to say it's forbidden, but you have to balance and not to be only ‑‑ and it was the three‑day in June before the ‑‑ with this forum with preference AI for good, okay, where everything is nice in the Geneva, during many decade there was a restaurant there was a black board saying two more rolls ‑‑ is free. 

Sometimes I think we are exactly in this situation.  So that mean we should be a bit more visionary, more critical.  And not to do the same mistake as for the security counsel.  Sorry to say that in this building.  I am from Geneva.  But the best country selling arms are those with the ‑‑ to keep the peace.  So don't do the same mistake a second time for robot, AI, and so on.  I think we have already ‑‑ here.  If you hike to use AI or ICT in any direction for care, for education, for security, for everything, okay, you have the 17 ‑‑.  So show that what you propose as the usage of the technology.  Resist is not worse by looking about the 17 ‑‑ okay?  And I think if you can declare that we use ‑‑ we should be with a lot of attention around human digital right and ability I think it's a key issue for the future and the future just now.

Yet I also have the feeling based on our previous conversations that the next wave of AI based technologies is different from previous technologies.  And you made already a comparison to internet technology and where you see differences.  And Karen maybe I can ask you again because you also in your previous comment highlighted and replaced the term artificial intelligence and started to zoom in on the question of autonomous systems and other panelists mentioned also what that may mean in terms of shifts of responsibility. 

We also talked about the scale, and someone also mentioned the speed of transformation that is ahead.  Can we have one round, what may be different about AI compared to previous technologies and what might it mean for our dealing with that technology?  Okay.

Way.  And I think that's where it's sort of ‑‑ a lot of the concerns, you know, is a speed and a scale.

So then there are different other things about what we need to do about it.  Whether we're going to stifle invasion because the invasion is going to bring a lot of opportunity so we should not do that.  On the other hand, we should be very careful how we apply it because we don't want us to turn into something which is very different.  Tim cook some sometime ago mentioned that I'm not worried about machines thinking like humans.  I'm worried about humans start thinking like machines.  So we should preserve that humanity to us, whatever we do, otherwise, we will be replaced by something else.

In contrast to natural intelligence displayed by the human beings, it's applied when the machine may make the cognitive functions that the humans associate with other human minds, such as learning and problem solving.  So I think that is the major difference between AI and other technological revolution.  Because the human beings has been the major actor in our society.  But if artificial intelligence can also communicate with the human beings and maybe in some areas the artificial intelligence can also be an actor in the governance system.  So that will cause a lot of problem. 

Maybe in some fields, it's strictly forbidden to those artificial intelligence.  But for most of the areas that artificial intelligence could be applied, I think that's the the problem for the artificial governance.  That's my point.

And that creates a lot of fear, moral panic, and suspicion, that this technology is going to replace us, deny us certain things.  And I think there's a fundamental misunderstanding of where the technology is right now, and also an underestimation of where the technology could be in the future.  And I think that that kind of uncertainty gap creates a lot of panic in that sense.  I think that artificial intelligence is different because it is designed to think based on human intelligence only in different ways. 

If we look at alpha zero, that devised incredible moves in chess that we had never thought about, this is something that scares us because here's a machine that was designed based on human intelligence but gave artificial intelligence.  However, this is very rare and very advanced.  And I think that the really good thing about having this panel, IGF, is that we can really combine where the technical developments are with the concerns.  And if we have that kind of conversation, I think that there would be less room for moral panic and more room for instructive and inclusive discussion.