Werner, it's great having you on #digitalduty and my blog here. You are one of the leading experts in this field. Let's talk about artificial intelligence. What's the difference between artificial intelligence and human intelligence?
Werner Vogels: I think the biggest difference is that artificial intelligence is just computer science. It lacks any form of creativity. It lacks any form of sort of the great things that make humans great. Artificial intelligence is just using data, some intelligence algorithms to make prediction about the future or to investigate things that we are doing.
I think intelligence is sort of greatly overrated in the kind of things, definitely where we are at this moment with artificial intelligence. I think there is two practical applications of this: I think machine learning is basically taking data from the past, sort of the emails that you send out to your customers who have responded to it and are based on that data, sort of knowing who you should be targeting in the future. Or, let's say, natural language processing. I think that is the other area. That is two very practical areas that now are being sort of seen as being majorly sort of mystic, but we have been doing this forever already.
If you have been an Amazon.de customer maybe, then you have received our recommendations forever. That was machine learning. There isn't almost anything within Amazon that is actually not artificial intelligence or machine learning, whether it is inventory level setting, whether it's fraud protection, abusive review detection. All of these kinds of things are what we now call artificial intelligence. But it's just computer science.
So, it will take time that we see the Turing test being passed or that computers start dreaming?
Werner Vogels: I don't think the dreaming is any near, apparent near. By the way, who cares? I mean dreams for us are important, because they are part of our creative process. Now it allows us to process things in moments that we are being not busy with sort of processing signals. And such a dreaming is a very important part of the creative process. But I think we are very far from actually sort of implementing this kind of creative processes in terms of algorithms. If you look at where we are, for example, with image recognition. If we encounter a new type of animal that we have never seen before, we can still figure out: Oh, that must be an animal. A computer, if it hasn't seen it before, it will never know what it is.
We've seen exponential development of data. We've seen exponential development of actors and sensors in the net. Do we, could we expect an exponential increase of cyber crime and cyber risk?
Werner Vogels: I think we have come forward in an early phase where people have not been thinking about it. I think the next wave of devices that are coming out, I think security needs to be the number one concern.
At Amazon we have been developing technologies, for example, that can be integrated in each of these devices. We call it Greengrass. Basically some of our secure cloud computing technologies can now be delivered on each of those devices, very strongly controlling who has access to them and what kind of communication they can have.
So, I think the early mishaps that we have seen around this world clearly can be protected. On the other hand, I'm also a strong believer in that security needs to be end-to-end. It cannot only be the infrastructure that protects our customers. We really need to make sure that as close to the customer as possible you need to put in protections. You have to be secure in an insecure world. And that can only happen if you provide end-to-end, let's say, security services.
If the infrastructure needs to be secure before you can be secure, that will never happen.
It's part of your duty. It is said Alexa has witnessed a murder. This causes an ethical dilemma: Which is the greater value? Privacy or the fight against crime? And how are you going to handle this issue in the future?
Werner Vogels: Well, first of all, I think that is a gross misrepresentation of what happened there. And so, no comment on that particular side of it.
I think privacy of our customers will be forever a number one priority, and so also the protection of our customers. And as such, I think, as we have talked about earlier with encryption, if you do good things for good people, it's obvious that sort of bad actors will make use of the same technologies. But that has been the case forever. It shouldn't prevent us from actually really supporting the privacy of good people to be able to really protect them.
Additional question – not in the video:
TechCrunch reported recently that Amazon would have acquired the cyber security company Harvest.ai. How important are security and encryption assets for Amazon?
Werner Vogels: I think protecting our customers will be forever a number one priority. It has ever been and will ever be. By the way, I think that anybody that does business online should make that their first, number one priority. Otherwise you have no business.
I agree.
Werner Vogels: I think we need make sure that security is built in from the ground up on day one. This can no longer be an afterthought. I think if we look at sort of the major vulnerabilities that we have seen in the past years, it's mostly been by old systems that have become network enabled and where we have seen sort of vulnerabilities in, because security was an afterthought.
I think there is a lot of education to do. There is a lot of technology to develop. And I think crucial and central in the whole place will be encryption. I think we are very strong believers in that encryption should have no backdoors. If encryption has backdoors, you might as well not encrypt. It's a bit like having someone giving someone the key to all your houses. Everybody will want that key.
