clean-IT: Towards Sustainable Digital Technologies

1. 01:18Yes, I would like to welcome you.
2. 01:20Finally a lecture hall where you can look into faces again, because those were bad times, of course we have lecture
3. 01:30continued.
4. 01:31At the very worst times the lecture theatre was empty, and that's strange, Because of course when you're talking, in interaction
5. 01:39you have to look in the eyes, it's difficult.
6. 01:41With the crowds you weren't, do people look at you smiling or are they quite horrified by what you say?
7. 01:47So we're glad it worked out, and we're glad that you accepted the, that you accepted the invitation.
8. 01:54I would like to welcome State Secretary Franziska Doktor Franziska Brantner from the Ministry of Economic Affairs.
9. 02:03and Climate I would like to welcome
10. 02:06Prof Dirk Messner, President of the Umweltbundesamtes, of the Umweltbundesamtes so only
11. 02:15and Mr Oliver Sühne, the Chairman of the Board and CEO of the Ecco Association and Dr Bernhard Rohleder, Managing Director
12. 02:28of Bitkom, we want to
13. 02:32think about clean IT and see what computer science, what digital technologies have to do with climate issues, I know
14. 02:43you have to do the arrow to point to this little person.
15. 02:50So, I think we are facing the challenges of digital transformation in our society.
16. 02:58many people say that the he has been talking about digitalisation for ages, it's old-fashioned, but what does that mean?
17. 03:04our society in all areas, I think we will find out yet and that are many profound changes.
18. 03:11than we have now in the first year.
19. 03:14That's one side of the challenge that humanity faces, this overcoming and when we have these big SDGs
20. 03:23that is the Sustainable Development Goals so the ones set by the UN as development goals, challenges,
21. 03:32that humanity is facing, these are the 17, they can see them.
22. 03:37Then it is the case that for almost all of these goals, digitisation offers an indispensable tool.
23. 03:46So to achieve those definitely of the 17, 13 will not be achieved without the IT technologies without the digitalisation
24. 03:57progress.
25. 03:59And that, of course, is a challenge.
26. 04:01It's not on one side that the technologies are developing.
27. 04:04We see that it offers quite great opportunities and to do things that are really essential to do, education,
28. 04:15fight against poverty, improve health and all these issues.
29. 04:19And by the way, HPI joined an initiative so that the Universities have such a dashboard, where on the individual objectives
30. 04:31the activities at which that university contributes.
31. 04:36And you can see from the colouring how strong that is, and of course that's to make visible a little bit on insight what it's
32. 04:44activities that are being carried out here at this university in the sense of achieving the goals.
33. 04:54will see.
34. 04:55That's quite good with us already with the colouring, that's what I said.
35. 05:01Great challenges, digitalisation, great opportunities, but now this is coming.
36. 05:06But digitisation itself of course has huge impacts on the environment, on the climate, on global warming.
37. 05:16The CO2 footprint of digitalisation is twice as big.
38. 05:22Today is twice as large as the CO2 footprint of all air travel of global air travel.
39. 05:30And they are always considered a bit evil not.
40. 05:32Then he came up with terms like flight shaming.
41. 05:35So we are very worried that there will be something like digitalisation, shame in the discussion, because digitalisation will be
42. 05:43is an indispensable tool for many issues in our society.
43. 05:48But we have to keep an eye on that, and what is happening is not that we can lie back contentedly and that
44. 05:55digitisation is the way it is now, but that it is exploding.
45. 06:00You see, just now we were talking about gigabytes and work, soon it will be Petabyte, Exarbytes, Zetabytes not imaginable anyway.
46. 06:09I don't know what new data will be generated and brought in the next few years through this digitalisation, but here are two
47. 06:18also very concrete figures, AI yes, on everyone's lips and great expectations of what can be achieved with AI in all areas of life
48. 06:29so one anyway with the modern technology that is now in development, in research.
49. 06:37If I train such an AI model, then you know, these are the ones that the Deep Learning.
50. 06:44These are such neuronal networks with hundreds of thousands of nodes and connection neurons, where very simple computational tasks
51. 06:53are executed.
52. 06:54And when you train them, when you show them a million pictures and always explain what it is to see on the picture and then
53. 07:02he shows them the unique and the first picture without explanation, then these networks already come around with good explanation
54. 07:10the corner, so partly better than humans in certain areas, which is quite difficult for machines.
55. 07:18The meaning of an image for machines are, This will be thousands, tens of thousands, hundreds of thousands of pixels, black
56. 07:25yellow, green, and something, and from that now to make out an image and a content.
57. 07:31This is a huge achievement that seems to come quite naturally to us, that we do every day, that we want to do this.
58. 07:37the ones that radiate, the signals that arrive on the retina as in In pictures someone we see what it transpose
59. 07:45so to train such an AI model requires the same amount of energy as 300 flights.
60. 07:55New York San Francisco with study 300 flights New York or five years in the life cycle of a car production, petrol, driving,
61. 08:09all that is produced there.
62. 08:10To train an AI model, that's a very complex model images later on you can always use that, but
63. 08:20such trainings are quite essential and that is somehow obvious, that will not work with rolling out, if we do that
64. 08:27roll out.
65. 08:27If we now assume that in every application, in every company there are now such AI models I'm going to be built in.
66. 08:37then it goes completely wrong with climate and global warming.
67. 08:42Second example Fortunately, word has already got around about the Bitcoin, which was a bit like this.
68. 08:48a saviour, not
69. 08:51Thing about later.
70. 08:52The less you understand the technology, the greater the expectation of what's solved, that's anyway, like
71. 08:59Bitcoin is operated today is in a form that clearly cannot continue, namely with these block
72. 09:08creation, which would then be recognised as worthy of trust by all, even those who do not trust each other?
73. 09:14Complex mathematical problems have to be solved with a huge energy demand and CO2 emissions of 22 million tonnes.
74. 09:23at this Bitkom mining and that's kind of the annual balance of a country again Jordan,
75. 09:32so digital technologies, problem or solution, that's exactly what our clean-IT initiative is supposed to be about.
76. 09:43not to give up and say it's too problematic.
77. 09:46Must look for something else, but just to see the weak point and to see the task what we can do in the
78. 09:55development of digital technologies in order to find the solutions that digital technologies can offer.
79. 10:03to solve the problems of mankind, i.e. CO2 emissions must be reduced.
80. 10:11Small remark now you sometimes also hear from the very big tech giants that they are already climate neutral, that
81. 10:19is of course a bit of window dressing.
82. 10:21It's not they paid, they really only get green electricity, but there's only a certain amount of green electricity.
83. 10:28If they use that, the others use the other electricity, because that's this mixture.
84. 10:33So it's not that easy to solve the problem, then I can say here, I have made an effort, we have here
85. 10:38has also green electricity but is not comparable to tech giants and the consumption they have so it really has to be
86. 10:45about making the IT systems and the software better.
87. 10:50That's just the way it is.
88. 10:50Plattner, who with the founding of the Hasso Plattner Institute a good 20 years ago wanted to create the possibility to work in this
89. 10:58field of digital technologies
90. 11:02university centre of excellence to build up IT engineers to support this digitisation, this digital transformation.
91. 11:10to help shape it.
92. 11:11And insofar we also see ourselves a bit in this duty to think about such issues, when others blame with
93. 11:20What have you done to our world, with your fucking digitalisation?
94. 11:24So, we started this Clean IT initiative in Davos 2 20 just before we started with Corona.
95. 11:35There is, that sounds similar, a Green IT movement, that's older, that's like 15, 20 years old, and that's focusing
96. 11:46on how to run data centres in a climate-friendly way.
97. 11:53With the data centres it is mainly about what happens with the heat that can be used to the second and the like.
98. 12:00but we are asking deeper and more immediately what do the it systems themselves consume?
99. 12:07Not only we have to see how they are and have to try to make the framework good, but you can ask
100. 12:12not consider this more when developing complexes using software programmes of ki solutions?
101. 12:20What about the energy requirement?
102. 12:24So the facets are as a university institute of course clearly to raise awareness would be to create it for
103. 12:33this problem, especially the circle of people who can really do something about it, the IT developers,
104. 12:41the IT architects who are building this system.
105. 12:46Second is the question if we don't develop models if we develop such blockchain activities if we develop software
106. 12:55programme in mind, what about the energy balance can't we achieve the same goal with
107. 13:04a lower use of energy and what we want to propagate a little bit as a solution the topic of sustainability by
108. 13:15design, so we don't build first and think about how to get it clean, but that's also an approach in the field of energy efficiency.
109. 13:24of safety.
110. 13:25Also pryvacy by design also because it is always very difficult to just modify the systems afterwards, so the T Mobile
111. 13:33T with you are its biggest challenge.
112. 13:37If we now see such a software programme in development, that is not easy at all for the developer.
113. 13:44or possible to perceive the energy consumption of what he develops, software then runs on machines that need energy.
114. 13:53Operating systems Technologie stack and then at the top comes the application programme where a developer architect of the feel
115. 14:03get it?
116. 14:04Is that a good solution, a bad solution.
117. 14:06That needs to be worked on, that is now we need measuring possibilities where you can try things out.
118. 14:13Say that is better why you can namely in algorithm.
119. 14:16To solve a problem there are many algorithms and so far they are looking at what it takes the fastest what it takes the least.
120. 14:24memory.
121. 14:25The question What needs least
122. 14:27Energy has not been on the agenda at all so far and that is actually an equally justified an equally justified
123. 14:34criterion for assessing which algorithm is good and solving the problem.
124. 14:39There is the next stage The algorithm can be implemented in quite different programmes, Even there have
125. 14:47we again have the whole world that the different implementations bring different energy requirements so the
126. 14:55is issue where we need to move forward in research.
127. 14:58There is no conference that is dedicated to the topic, there is no scientific journal that is dedicated to the topic.
128. 15:04that cares about it, that catches the scientific community, has to start to think about it and to take that into account in order to
129. 15:13move forward on that.
130. 15:15I'll give a small example this afternoon we'll show more and discuss more solutions, what you can really do concretely.
131. 15:22can do.
132. 15:23In my research group, the idea is that we want to address this issue with this network with the AI networks
133. 15:31And there it is that the Deep Learning, the great AI technology, has brought AI back into the conversation in such a way that this is connected with this
134. 15:40huge networks that works in this neurons with 32 bit architectures there will be, so Convolution, a relatively simple
135. 15:49Operation multiply Whatever you what That's done on the basis of 32 bit architectures.
136. 15:55The question was Can't we just do it at the one bit level, and when you do that, directly do it without a big thinking
137. 16:03If you do it directly without thinking about it, then you save a lot of energy, but the accuracy, that's the measure.
138. 16:12How good are they in their predictions, how good are they in the in the in the solution determination they are clearly worse
139. 16:21than those of the 32 bit architecture over 10% is this accurate heiz biz and the competition in developing such
140. 16:30AI system is How accurate are you in your solutions answers so the question How can we work on this?
141. 16:38So directly translate is not enough, but then there must be new architectures, new approaches to get that done.
142. 16:45At the moment we have made good progress, so 95 % energy demand and we are now only at 5 % loss.
143. 16:55On accuracy In some issues it's not that bad, but it has to get better somehow, that's the
144. 17:03Challenges in research, but also affects other topics How is that?
145. 17:08Data storage is a very energy consuming process, data will continue to explode, rethink with each new application
146. 17:16Internet of Things.
147. 17:17If something like that comes, it's not going to be a multiplication, it's going to be an explosion.
148. 17:21This exponential growth in data are how do we go around with that, how is that with data centres and cloud that are now
149. 17:29new this computing that once again things We're still talking about going to the cloud, but in the research always there
150. 17:36thinking again, which parts you don't need to put in the cloud at all, but you can solve on-site much better
151. 17:41to then only give the results into the cloud, so that you can also do that in the end, for example, with Clients with small
152. 17:48devices.
153. 17:49Smartphone or similar streaming is a big issue.
154. 17:54Of course, everyone in the world can watch their own movie version of the same movie seconds apart, but can you do that
155. 18:00not make it smarter that I then wait five seconds and then somehow have it in the pool which is then streamed
156. 18:07that people are looking at at the same time, so that I don't have this amount of data huge amount of data at the same time
157. 18:13at the same time?
158. 18:14At blockchain, a good technology is only in the form that it used to be, that is, how it was developed, there it is
159. 18:22too energy consuming and of course much more.
160. 18:24Especially if we think about the Internet of Things, so this Clean IT initiative was such an idea, so that bit to talk about
161. 18:31and we were also very pleased that we found comrades-in-arms, immediately SAP, the ECO Association
162. 18:42Acatech, also universities such as the Technion in Israel, with which we are connected through the Research School and that is
163. 18:50very important for the implementation later on.
164. 18:53IEEE is the world's largest engineering association.
165. 18:58So the standards sets about standardisation, if you know exactly what screws you can turn, of course.
166. 19:05through standardisation then suddenly bring that into broad effect?
167. 19:10We have tried with the means of openHPI, that is our platform, I hope all of you know the very strong uses of openHPI.
168. 19:20over a million 1.2 million enrolled learners, where so much is offered to explain digital technologies
169. 19:30we have also set up a channel where we invite everyone, a clean it forum where we invite everyone.
170. 19:38Their best project is to also put their ideas there for publication in the public domain.
171. 19:45There is a monthly colloquium, that is good, we have also found a hearing in politics, so now in this one.
172. 19:55year the G 20 is under Indonesian presidency there is a paper a clean it paper in the political discussion
173. 20:05so there the question you can't think about developing labels than first it would be research and such issues
174. 20:12promote label to develop for good, especially good solution.
175. 20:18And also in the G seven process, which is under German presidency, this issue is now being discussed.
176. 20:28So with our conference clean-IT conference the first, they also want the continuation.
177. 20:34Do we want to create such a platform, because we see efforts in this direction in many places, but these efforts are not being continued?
178. 20:40all of them are not perceived.
179. 20:41If you bring that together a bit like that, then you can make a lot more school with it than that is the case so far
180. 20:49is to bring into consciousness than it is a very optimistic view that through research through technology in these technologies we can
181. 20:58much better, cleaner, so don't slacken in research, but push it forward, just with especially
182. 21:05so and here is the conference, the procedure.
183. 21:08We have the whole day planned.
184. 21:11I warmly welcome the various speakers who are then always introduced.
185. 21:17We start with a keynote by Mrs. State Secretary Brandner, we then have a 1st panel.
186. 21:28I think they all have this our booklet, our booklet lying in front of them, which is a bit more than simple
187. 21:36Conference booklet, which is supposed to give a bit of information on the subject as well.
188. 21:40We have partners in this conference.
189. 21:43SAP WMWare Huawei, the EU You can see that we thank everyone very much.
190. 21:52We have changes in the, in the, in the lecturers, in the presentations.
191. 22:00Especially him.
192. 22:02Doctor Herzig from SAP, who is responsible for a large area and is presenting, is Corona infected from India at the weekend.
193. 22:12but will be replaced.
194. 22:15Mr. Heitkötter will present the work.
195. 22:20So I'm glad that we're finding support there and today hopefully an instructive and instructive doesn't mean
196. 22:32that it can't also be a joy to come together and that what has always been missing from the video conferences, which is that
197. 22:41you can stand out there together over coffee or tea or juice and also talk about things that are not on the agenda.
198. 22:48that are not on the agenda, happen to come up and can network through that.
199. 22:52Let's be happy that this can be a real live event again.

To enable the transcript, please select a language in the video player settings menu.