00:00:00: impulses and perspectives the DIKT expert talk hosted by Nikolai behr hello

00:00:09: and welcome to a new edition of impulses and perspectives today I'm talking to

00:00:15: Richard Polivka he is an active scientist at the CERN in Geneva or near

00:00:21: Geneva and he's founder of AI consulting hi Richard Nicolai thank you very much

00:00:28: for the invitation so the corona crisis brought a lot of

00:00:33: new things people work in home office students do a big leap into digitization

00:00:40: they have to do homework now online and so all the world is talking about AI big

00:00:48: data and maybe also the solution for the corona crisis will be connected to AI so

00:00:56: is that a good way to find like a vaccine or a medicine for the corona

00:01:02: virus or the covet 19 virus absolutely you are you're spot-on actually as a

00:01:09: matter of fact it was an AI company called blue dot which already in 31st of

00:01:16: December 2019 first time reported that they see something like outbreak coming

00:01:23: because what they do with the help of AI they are analyzing multi multiple

00:01:28: sources of text and website publications government releases etc and they can

00:01:36: they could pull all the different information together and they were

00:01:40: starting to see patterns how that something is coming so that was already

00:01:45: the very first thing but you are absolutely right the AI which is able to

00:01:51: with the strength of AI is analyzing this huge huge amounts of data which is

00:01:56: really not possible for a human and finding patterns and the hidden

00:02:01: correlations which also for us which would be very difficult to spot so it's

00:02:06: a natural candidate to use really and it was also done for example for the genome

00:02:12: sequencing the virus that was done with the help of

00:02:15: AI and then of course the next step is to find the right vaccinations so

00:02:20: typically you could do this in two ways you could sort of program in the

00:02:26: chemical reactions you might think are the important ones and then you try all

00:02:33: kind of possibilities because that's as well where AI is very good like

00:02:37: repetitive tasks with just a little small change of the parameters to really

00:02:42: explore all possible solutions or actually there are techniques called

00:02:47: natural language processing and they're basically there are algorithms which can

00:02:53: scan through all world's database on whatever topic you would like but in

00:03:01: this case obviously it would be very low G and the commad 19 so these algorithms

00:03:07: can scan this and can propose what has been done before and worked and maybe it

00:03:12: could be tried this time so this is the way how a I actually can can help with

00:03:18: the development of the vaccination so it's more or less a matter of speed and

00:03:24: variation so the AI can do things faster and then a big more feel of varieties

00:03:34: than the human and who is in the lead there in the pharmaceutical industry for

00:03:42: example there are many new companies which are basically the whole world is

00:03:49: now trying to really strive for the same goal and I would say it's a company

00:03:56: competition between various companies or providers so I wouldn't dare to really

00:04:02: name one outstanding but I would say either if you have a pharmaceutical

00:04:08: company which is already large enough they must have by now recognized the

00:04:13: power of AI in drug development because that's that's by the that's per se not

00:04:19: the new tap topic so the big players typically already have this that in the

00:04:25: house and can use this enhance their potential to find the

00:04:29: vaccine or there is a new sort of way of approaching this which is really a

00:04:35: technological startup with some of subject matter experts in the team who

00:04:41: would be really targeting going after the solution for the vaccination and

00:04:48: then of course you have all the university research which is I believe

00:04:51: combined in the global effort so you have scientists you have institutions

00:05:00: like the CERN you have universities doing the research but on the other hand

00:05:05: you also have the industry and and sometimes the business people do not

00:05:10: talk very well to the scientists and that's I think is the reason you founded

00:05:15: AI consulting to bridge scientists and the business world

00:05:19: what are your experience in how do they work together usually so just one side

00:05:28: note for example institutions like CERN who the primary focus is the fundamental

00:05:35: research in the realm of physics we are contributing as well to the kovat 19

00:05:42: vaccination hunt but only in the way of putting available parts of our huge

00:05:51: computing farms which we normally use for physics but of course now the health

00:05:56: has priority so we have allocated a lot of space which then sort of computer

00:06:01: heavy programs of the actual sort of expert a ion vaccination finding then

00:06:08: can use so that's just a side note and to your main answer yes these two worlds

00:06:15: are very different but it's much more on the level of communication actually and

00:06:23: learning to speaking the right language because the physicists typically being

00:06:30: in academia they have a way of thinking they have a way of talking about

00:06:35: problems but then if you would then put straight away in a business consultancy

00:06:40: that not necessarily work well but that's why

00:06:43: I started AI consulting where we already know because my own expertise from my

00:06:51: own expertise I know what type of problems we are good at solving with

00:06:56: sort of the team of some of the brightest minds from in the world which

00:07:00: work at CERN so I know the class of problems we typically can solve and I

00:07:05: did the 400 analysis of the market to see where we could easily apply these

00:07:10: talents and way where we could actually be beneficial for the customers is that

00:07:17: in a certain field of science so more in physics or is that also for biochemistry

00:07:24: or other fields it's in a specific niche but it's not sort of biologically

00:07:31: oriented it's more in the financing sector so we are developing risk

00:07:39: assessments we are developing automated algorithms how to credit score how to

00:07:46: evaluate loans and and this kind of the most typical projects we are solving

00:07:51: however its due to the fact that we have a strong connection with the subject

00:07:58: matter experts in that particular field so of course if we find a person who has

00:08:06: very strong and interesting idea how big data and complex mathematical algorithms

00:08:13: could help let's say in biochemistry there is always a way how to how to have

00:08:19: a discussion and see if there could be a way how to work together because we are

00:08:25: always open for new and interesting challenges so what are the biggest

00:08:30: drivers for AI so we talk now about financial institutions about physics

00:08:39: about pharmaceutical companies is also the de car industry with the autonomous

00:08:46: car a big drive of AI of course self-driving cars computer vision right

00:08:53: that goes hand-in-hand and I think every one

00:08:56: of us has used AI at least once a day like just just take this right now it's

00:09:03: of course not so much easy to to use but normally when you look for a path to

00:09:09: work or to holidays in Google Maps that's all a I as well so there are many

00:09:15: industries like you said the financing for example or the self autonomous

00:09:21: driving which where the AI is already well established heavy industry heavy

00:09:27: machinery traffic you know for example trains planes there is there is a very

00:09:35: big area where artificial intelligence is used and that's called predictive

00:09:39: maintenance because with all the sensors for example a train can have you have

00:09:45: many early indicators that may me a carriage needs to be replaced or that

00:09:51: something needs to be fixed and it's much better for a company let's say like

00:09:56: Deutsche Bahn to implement AI into the into the systems which can say you

00:10:02: actually don't need to put this train every month for one day of stop for a

00:10:08: general check you just wait for the sensors to tell you oh actually now you

00:10:14: should replace that now you should replace this so that's been going on

00:10:18: already since some time and it's quite powerful well I would hope that what

00:10:24: this whole coronal crisis could improve is actually the usage of AI in health in

00:10:31: the health care industry and I'm also connected to the AI for good initiative

00:10:38: which is driven by the United Nations here in Geneva and they have a very

00:10:42: interesting series of webinars on how a I can do good things so for example

00:10:51: there was a and that's very relevant for the kovat 19 where you actually should

00:10:57: stay home you should avoid contact so what's becoming more and more

00:11:01: important is to be able to self diagnose yourself and it's different than if I'm

00:11:07: just googling myself oh I I'm sneezing and I

00:11:12: am losing taste then if I have actually in my smartphone specialized app which

00:11:18: can have basically a conversation with me and that's already won AI application

00:11:24: which is called chat bot which you may have experienced as well just by talking

00:11:29: trying to get an information from your bank they typically use chat BOTS

00:11:34: well even though sometimes you think that it's a human it's usually more or

00:11:39: less smart chat BOTS so having a chat bot in your a I phone application which

00:11:44: could guide you through self diagnosis and then send this information to the

00:11:50: database to another AI which is in which would be sitting then at the hospital

00:11:56: server which could then sort of compare see how how your profile compares to the

00:12:03: rest of the city and so on make a thorough analysis and could evaluate how

00:12:07: serious basically your case is you know and that could help a lot by just a

00:12:15: tracking the population for the spread but be also removing some traffic of the

00:12:23: phone helplines I don't know if you try to make a call

00:12:26: again to a bank it was my case yesterday I spent 30 minutes listening to music

00:12:32: before I managed to get through so if you manage in crisis situation to

00:12:36: offload the traffic from the common ways that's that's an improvement and taking

00:12:44: it one step further for example if you have this kind of self diagnosis

00:12:49: application and you think not Geneva but you could think some remote places in

00:12:55: the world like where it's very difficult to get any medical help you could see

00:13:01: that you could combine this with an autonomously flying drone which when it

00:13:06: gets sort of the order quote-unquote from the self diagnosis it picks up the

00:13:11: right medicine and it flies 20 50 kilometres to the village and then

00:13:15: delivers the medicine there and comes back it's completely contactless you

00:13:20: know of course standards needs to be developed and

00:13:24: especially the self diagnosis accuracy needs to be very high but I believe that

00:13:32: that's a good way to go and we will see things like this in the future so we

00:13:38: have now the the public debate on the corona app and especially in Germany

00:13:43: people are very critical if someone uses their data it's kind of strange because

00:13:52: whenever there is a some action of a company where they can win something

00:13:58: some lottery they give in all their data but if they should give in their data on

00:14:06: on a freebase they're always afraid who can use this data and isn't it that

00:14:15: dangerous so is scientists talking enough about AI and the positive results

00:14:25: and benefits for the people or why are people still so afraid of the

00:14:32: digitization and the misuse of data it's funny then you put it this way because

00:14:40: I'm sure that Facebook knows about me way more than I would ever put into any

00:14:45: governmental app but it may be a little bit psychological in the sense of like

00:14:51: you know of this George Orwell style scenarios of having an government sort

00:14:59: of knowing everything about you just tracking wherever you go and so on

00:15:03: but it seems too much science fiction movies but sometimes you see the future

00:15:10: there in terms of the technological advancement just take the famous Star

00:15:15: Trek from the 60s we have quite a few gadgets which they were using they're a

00:15:20: science-fiction they are already reality now but what I wanted to say is and I

00:15:27: know that from firsthand experience when you really want to develop an app or

00:15:33: functionality like this there so many legal aspects of this and sort

00:15:42: of cybersecurity aspects of this then if he wouldn't pass the tightest criteria

00:15:48: on these things you wouldn't be able even to do this type of apps or business

00:15:54: so I would really say regarding the amount of data there is and the misuse

00:16:01: ability of those people should be much more careful about what they are putting

00:16:06: on Facebook then that there is a corona app which I would say it potentially has

00:16:12: a good chance to really track I think it's usually called something like smart

00:16:18: smart quarantine or something like this really be able to track who is who has

00:16:24: come to the proximity of someone who has the virus and so on and be able to

00:16:29: really see and contain the spread so I think that's quite useful what would

00:16:33: need to be assured and I think that's the key is really find a way how to make

00:16:43: sure that the people have the ability and they trust that when let's say after

00:16:49: the corona they hit the delete button really the the information gets deleted

00:16:55: from all servers where it was and so it's similar like the gdpr which is

00:17:00: doing a lot in terms of the data protection so from this perspective if

00:17:05: the app is sort of gdpr compliant for example it's the the highest level of

00:17:10: data protection security you can get in this moment people are afraid that they

00:17:18: that they are controlled by governments by insurance companies and so on but

00:17:24: from a scientifical point and and and we just talked about the necessity of such

00:17:30: an airport that the positive results it could have or should have to find out

00:17:35: who could be the next corona victim and what is is it worth taking the risk to

00:17:44: give your data into this yep yes I think it is worth doing it and

00:17:51: it just needs to be carefully manufactured by whoever is responsible

00:17:59: which should be probably the government in terms of showing people that they are

00:18:04: not collecting more that people already publish anyway you know you put the

00:18:10: example of the insurance okay that that's a good one but what if the

00:18:14: insurance gets access to your Facebook page where you are putting pictures how

00:18:19: you are jumping of a cliff or how you are done you know skiing in the

00:18:25: mountains out on the piece or smoking or whatever so you really it's much more of

00:18:33: a I would say marketing campaign from the government explaining that they are

00:18:37: not requesting any new information this information is already there it's just

00:18:42: for this particular purpose and it has it of course it has to have all the

00:18:48: state of the art ways of securing the cybersecurity itself because it's two

00:18:55: different things one is misuse of the data you collect and the other one but

00:18:59: by let's say consent of the person who is issuing the app and the second one

00:19:04: would be that just someone hacks the application right what fields of our

00:19:10: daily life will change through AI within the next years many things like as I

00:19:18: said before the AI is already used in many of the industries there are new

00:19:23: coming I would as I said I would strongly hope that sort of AI enhanced

00:19:30: health system you know our augmented doctors and this kind of applications

00:19:37: could help and basic because the best solution is always that you combine the

00:19:43: best of the both world the AI world and the human world and wherever you need

00:19:48: really like some heavy lifting of data and and pattern recognition and so on

00:19:53: you use the you use the AI and then when you need to use these outputs to

00:20:00: make an informed decision that's where and where the human comes in the AI

00:20:06: doesn't have the imagination so it's it's very broad fields I would say there

00:20:12: will be a rise in entertainment you know more more fancy Netflix's and these kind

00:20:20: of things because this is on the rise right now e-learning for sure maybe we

00:20:27: will experience some big changes of how we do how we educate our children I can

00:20:36: imagine that if many people and many companies realize that it's much more

00:20:41: comfortable and basically equally less equally same productive to work from

00:20:48: home the company realizes that they save a lot on the transportation right and

00:20:54: flying and then everything so that might have huge changes on the society itself

00:21:00: how far it's gonna go that's that's very difficult to to see another field which

00:21:09: is really well suited for a I sort of logistics so this problem like you have

00:21:15: a supply chain and you need to figure out how to how to deliver your goods

00:21:24: over even over Germany with let's say I don't know 80 cities it's quite a

00:21:29: non-trivial exercise now what I've seen that there are people talking about

00:21:38: effects something called D globalization maybe we will see less of like the

00:21:44: global supply chains you know that things get produced in China then

00:21:49: instead of transported here added more value here transport it there it may be

00:21:53: that countries will be much more careful where exactly are the all steps of the

00:22:01: production chain and we'll try to keep them closer to their place exactly for

00:22:05: the situation we've just experienced now process optimization and logistics

00:22:12: optimization that's big topic for AI and for complex

00:22:17: analytics so that's another way how I think AI will help sort of for shaping

00:22:22: the next year's will a I help us to solve the climate change problem or

00:22:32: let's ask me the other way around will there be a solution for the climate

00:22:37: change without AI I don't think so and it will help us if we want to AI is

00:22:45: always a tool and again

00:22:50: for example methodological forecasts those calculations are incredibly

00:22:56: complex right the models of weather it's it's chaotic almost so there is a lot of

00:23:05: complex problems in the environment which can be solved

00:23:10: sometimes even only through artificial intelligence think also like

00:23:17: distribution of resources in terms of water you know there are certainly areas

00:23:22: of earth where there is a big shortage of water already now or coming the more

00:23:28: data you can collect the better and then these days with the rise of this

00:23:33: Internet of Things you know every little device can be collecting all kinds of

00:23:38: data and sending it somewhere the more these kind of sensors you put to the

00:23:45: places where you really need to get the information from imagine like water dam

00:23:50: somewhere in the middle of Africa where it's very difficult to get to if you

00:23:55: have everything controlled remotely or monitored remotely you are much more in

00:24:01: sort of in control of what can potentially happen so I would say yes

00:24:07: it's definitely a - how to how to help to tackle the the climate change but

00:24:12: it's the decision to do it it's it's not on the AI scientists

00:24:19: thank you though much a Richard there was really great insights into AI into

00:24:26: the corona app the risks and the chances I wish you all the best stay healthy and

00:24:34: have a good time at CERN stay safe bye bye if you would like to watch all of

00:24:43: our new episodes please subscribe to our YouTube channel