Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
#11 - Paul Springer - AI for the common good
64 Plays7 months ago
Ever wondered how AI can be harnessed for the greater good? I talk to Dr. Paul Springer.
Paul has a background in theoretical physics and years of experience as a management and IT consultant. He’s also the co-founder and managing director of MI4People, a non-profit that develops open-source AI solutions for humanitarian, social, and environmental issues.
We dive into how AI can help tackle global challenges, the mission behind MI4People, and how technology can drive meaningful, positive change.
Enjoy!
Recommended
Transcript
Introduction and Guest Announcement
00:00:00
Speaker
Hey, what's up, everyone? This is Ajmal Savary and welcome to another podcast episode. Our guest today is Dr. Paul
AI's Role in Humanitarian and Environmental Issues
00:00:07
Speaker
Springer. In this conversation, we talk about AI for good, specifically how AI can be used to solve humanitarian, social and environmental problems and the hurdles that are in the way. Enjoy.
Podcast's Aim and Dr. Springer's Background
00:00:35
Speaker
Hey, everyone, and welcome to another podcast episode. If you're new here, my name's Ajmal. I'm a neuroscientist and entrepreneur. On this podcast, we explore the links between science, technology, business, and the impact they have on all of us.
00:00:50
Speaker
Today we talk to Dr. Paul Springer. After his PhD in theoretical physics, Paul worked as a management and IT consultant in the financial sector. He is a passionate AI enthusiast and entrepreneur. Among others, he co-founded Mi4People and leads it as volunteering managing director and chief scientific officer.
00:01:09
Speaker
Am I for People is a nonprofit organization
MI4People's Mission and Challenges
00:01:12
Speaker
that investigates how AI can be used to help solve humanitarian, social and environmental problems of this world and builds corresponding AI for good applications that are open source and free of charge. All right, enough background. Let's get into it, shall we?
00:01:30
Speaker
Welcome to another episode. Today we're talking to Dr. Paul Springer. Well, full disclosure, right? We met through Christian Helles and I'm on the ethics committee of MI for people, which stands for machine intelligence for people. And in an upcoming episode, we will also talk to Christian.
00:01:51
Speaker
But for now, let's talk about MI for people. So Paul, I mean, you got me ah into the ethics board because you know I've been in the AI field for a while, um more in the medical sector in that area. And that's also where MI for people is also doing things. But you know to get started, um let's talk about AI for good in general.
Defining 'AI for Good' and MI4People's Approach
00:02:17
Speaker
And because there's so much to talk about, I think one thing that surprised me a lot was how difficult it actually is to um you know to do these things, to do machine intelligence or AI projects for for good, um or let's say for the tech community to have it maybe a bit more open source. It's it's not as simple as to just press Git commit and you're done with it. um There's so many hurdles in the way.
00:02:47
Speaker
but before i talk your ear off um Let's dive in and start with a very few concrete things. Can you tell us a bit about AI for good? And maybe it's best to start with MI for people. Yeah, sure, Jamal. So um yes, I'm representing today MI for people, ah one of the organizations that I co-founded. And it is a nonprofit organization that is doing AI for good projects. so And for the listeners who never heard about I
00:03:19
Speaker
There are really different definitions of it, but I like the definition that says, okay, it's basically AI that was developed to help solve humanitarian, social or ecological problems. yeah ah And, um you know, AI is like transforming ah our daily lives, our jobs. And I think there is no human on the planet Earth who didn't get that AI is will have impact on him or her. um But the majority or basically everything what we heard about it is like kind of commercial AI. Is it help you in some manner, but usually it helps you you know to be more productive or to save money or be better in your skills and so on. ah But
00:04:13
Speaker
AI can do much more and in fact if applied correctly and if designed correctly it can rescue human lives, it can fight hunger, it can have fight poverty or it can ah fight ah or at least ah help us understand climate crisis, ah can help ah us ah clean the oceans from the garbage floating around and so on.
00:04:40
Speaker
These applications of AI there are, in my opinion, much more important because they have so large potential impact in the world. But in most cases, they have no business case. And it is the already the first hurdle for such project is where you get money from for it. Because if you have no business model, you cannot get money.
00:05:06
Speaker
cannot earn money, nobody will invest in it. yeah Even so, the impact might be extremely large, yeah almost thr revolutionizing partially. Yes, but in general, I forgot it's AI that helps solve really big problems. Usually it has no business case and it is why MI4 people exist. We are a network of AI experts, data scientists, machine learning engineers, cloud engineers, or whatever expertise expert people who are needed to create such AI for good applications end-to- end to end.
00:05:49
Speaker
And we are all volunteers and we work together in our free time to realize such applications in a structured manner. Okay, wow, that's a lot. I mean, you you talked about, um well, big problems and i'm I'm not necessarily talking about big business problems, but humanitarian problems even. um Can you start with just so so the listeners can get a better idea? Can you start going through one of these projects you mentioned? I mean, it can be environmental one or the med tech health, healthcare sector one.
Innovative Projects by MI4People
00:06:27
Speaker
I know there is no shortage of examples that you have um just so the listener can understand a bit better going through them. What, what the AI application
00:06:40
Speaker
actually does and how it can help organizations to yeah too make their process better or maybe even do do their work in the first place, get it done in the first place. Yeah. i say ah In fact, we are focusing more on projects that are not so much about optimizing processes, but more about opening completely new ways of solving problems or of generating knowledge. And so we currently have 10 projects running.
00:07:10
Speaker
And maybe in fact, let's start with environmental one. um um We are using ah AI and Pavlik satellite images. to spot marine litter in seas and oceans. So, basically relatively simple thing, you take a satellite image, have some kind of a segmentation model bro running on it, and it gives viewers information pixel by pixel. Is it ah some kind of garbage or not? With some confidence, of course, yeah. um And to um ah why we are doing this is ah because
00:07:50
Speaker
We know, we all know that our oceans ah are polluted extremely and we all know that it is bad for environment and it is bad also for us in the end. Microplastic, for example. all right right ah But also, you know, like, like um ah beaches full of garbage is also not nice. Yeah, it's also, ah for example, has negative impact of economy and ah the regions that ah rely on tourism, they get big problems. um Not only the companies, but the whole regions get big problems because of the ah garbage in disease um and And every now, every one knows that there is a problem, but
00:08:40
Speaker
Nobody knows first how much exactly ah ah garbage we have in the oceans. Everything you find in research papers are estimates that are not very precise. But the bigger problem is that you don't know where it is, maybe except of a couple of patches.
00:08:57
Speaker
ah <unk> really bad ones right yeah But in general, you don't know where it is because the situation is always changing because you have streams and oceans and the situation just changes all the time. And it means for those who want to collect garbage, it is difficult to plan their operations.
00:09:21
Speaker
And ah we started this project with the idea, OK, we want to help ah those guys, those nonprofits, ah and partially also government organizations that clean oceans and seas to do it more ah effectively, more productive with less cost and we also with less CO2 emissions. Because you know such operations require ships, and ships produce quite a lot of CO2 emissions.
00:09:50
Speaker
Yes, um and um we hope that with this application we can enable such organizations ah but also partially even individuals to contribute to gathering garbage from disease.
00:10:08
Speaker
ah more productive. yeah ah But this project has also another very interesting side effects that was not obvious for us since the beginning.
AI's Impact on Climate Research and Healthcare
00:10:17
Speaker
And it is about measuring ah ocean streams or sea streams.
00:10:24
Speaker
ah yeah so it's The interesting thing, so if you ah start to talk to oceanologists... By tracking the garbage movement. You can use the system basically tracking for tracking the garbage and calculate back the ocean stream. And you can do it with very high ah precision. and yeah so ah For example, streams in...
00:10:51
Speaker
in the Baltic Sea, here in the north of Germany, they are in fact white whale measured, but how they measured it, they used some pieces of wood And they just, you know, uh, like, um, let them at some, uh, the different locations and just let them flow. And then they were waiting on the shores and looking where it comes. I see. And then, uh, from the distribution, the calculated bags. as the ah the seeker Yeah. And it is like one of the most precise, uh,
00:11:36
Speaker
um techniques right now but it is like you cannot do it everywhere in the world because like oceans are really big yes but you have already everywhere this hypothetical ah pieces of food yeah it's garbage not garbage necessary or made of wood but you have it overall and you can use them to to to calculate the ocean streams to measure them and this is in fact extremely important for climate research ah because many climate models use ocean streams as input and if the streams are not measured precise then models are not precise so typical propagating error. yeah yes yeah um And also you can ah you know if you um
00:12:21
Speaker
and ah monitor to the ocean streams using these techniques overtime you can also measure the impact of climate change on the ocean streams with this again you know like.
00:12:32
Speaker
like ah the back coupling, yeah your climate change is changing, ah sorry, climate is changing, and then your your streams, it becomes different, what again, causes many different problems. And through our system, we hope that in the future, we can really monitor it, basically real time, yeah, the impact. ah ah So therefore, this project, in my opinion, it says like, very big,
00:13:02
Speaker
but war But it has potential for extremely large impact. yeah But again, for example, right now now, nobody would invest in it in profit oriented context. yeah so And it is why we need volunteers working on it.
00:13:19
Speaker
Wow. Okay. so i knew Of course, I knew already about this project, but I had no idea that climate scientists are using the same data to basically reverse engineer um the the models for the streams. Yeah. It was in fact kind of an anecdote how they came to this idea. It was like, I think the the first time this technique was ah used, it was some years ago, a big ship ah went on rifts.
00:13:49
Speaker
And as this ship was full of, you know, this ah yellow ah yellow rubber ducks. Ah, yes, this is basing rubber ducks. yeah and Yeah. And it just lost a lot of rubber ducks. And then people just started to collect them on shores in different countries. I think it happens somewhere in ah close to um Australia, I think. And I'm not sure. But then people start to observe where these ducks are found later. yeah And from the distribution of observation on different shores.
00:14:20
Speaker
the who would make a quite interesting statement about ocean streams. And now people start to do it in a more controlled environment.
00:14:31
Speaker
but you know steel so like this um With the garbage, it's basically the same idea, but much, much easier. And basically you can do it for for the whole world in a relatively short amount of time.
00:14:45
Speaker
Well, you know, for simplicity reasons, I would have expected they just, I don't know, have a little boat with a GPS sensor and throw it in the ocean. And that's basically how they do it. yeah yeah yeah but It's too expensive. Yeah, but in the end, it is too expensive because, you know, to map the whole world, you need a lot of such plots. Really a lot of such plots.
00:15:06
Speaker
Yes. And there's, of course, are also different methods. There are some some stationary measurement stations that observe current with those others. yeah um ah ah But generally, we have currently, I would say, no scalable option for measuring ocean streams. And this this application might become such a scalable approach.
00:15:33
Speaker
So trying to make something good out of a bad situation already. And I mean, later on, when when the cleanup actually happens, you can use other material that doesn't cause microplastic or any type of yeah substantial pollutions. Yeah, yeah that that's true too. And of course it is can be also used for other, as ah not only for gathering, it could be ah also used to spot the main sources where plastic is coming from.
00:16:03
Speaker
the region where the river, in many cases, there are rivers that are sources. And you know if you know the worst polluters, you can better um yeah create actions. It is also important for ah regulatory parties to make um more powerful and you know um a target um
00:16:33
Speaker
targeted policies and and measures so to prevent pollution, not only for cleaning but also for prevention.
00:16:45
Speaker
Okay, that's interesting. So a friend of mine, she's working at a and an AI startup in and of course Silicon Valley, right? i mean like else And what they are working on is um They are trying to use the technology to prevent um or early detect forest fires, which is I thought was, I never thought about that application, but of yeah, I mean, sure, why not? all right If it can help, it's great. But at the same time, it made me wonder,
00:17:21
Speaker
Well, who's going to pay for that? But of course, in the United States, things being a bit more privatized than in Europe, um they see, of course, a huge benefit in not having half of California being burned down by forest fires. On top of that, California is an economic powerhouse. So for them, spending a little bit on this is nothing.
00:17:47
Speaker
um But you mentioned that, I mean, compared to that story, the reason I'm bringing it up is the funding. They, of course, received funding for this. I mean, of course, and not saying that everything in Silicon Valley that receives funding has has a sound reason to exist.
00:18:04
Speaker
but
00:18:07
Speaker
They do have a business case, it seems. yeah yeah and it makes it's I find it so strange that here, that doesn't seem to be the case. You know, it is in fact a very, very interesting discussion. and Even in this example I brought to you with a marine leader, I can imagine that at some point it might be given in this case, but currently it's not a business. There is no business case. Or if we're talking about wildfires, yeah as you said, it is a business case in in Silicon Valley in ah in California, but it's probably where you have landlords trying to protect it or also governmental organizations.
00:18:52
Speaker
yes But let's speak about a forest in a relatively poor Southern country. yeah so You can use basically the same technology probably, but there is no no money to be earned and therefore the application is not realized for this region. Even so, these regions might be even more in danger.
00:19:13
Speaker
Yeah, it is like um um many of our medical projects and have this kind of situation. yeah so one our A medical project is about ah AI and radiology and maybe your your listeners who are into AI might love here a little bit because the AI can help here. It is a very prominent factor. I think it was like 2018-19 where
00:19:45
Speaker
People claimed that AI radiologists are better than human radiologists. yeah yeah And so already a while ago, and this technology is in fact already used in in US or in Western Europe or Europe in general.
00:20:01
Speaker
And there are startups and also big corporations having this kind of technologies and it is already kind of providing something good to people in this regions. Yeah, it is. It enables radiologists to be more productive to make better diagnosis and better decisions. But let's then think about the development, the developing world. Yeah. There are um um ah many many hospitals ah and we would in fact talk to them it's not like the assumption we talk to those people to say to us so
00:20:37
Speaker
We have this much X-ray machines. Somebody donated them to us, as they usually a bit older. yeah But we have no one radiologist who can make diagnosis, who can read these images basically in correct way. So in countries like Germany, and I'm sure also in US, ah such hospitals will be not allowed to make any statements about radiology or imagery. And in these countries, there's like uneducated general practitioners oh I think sometimes even nurses must make diagnosis because there is no better way. um But yeah they have to make do with what they have, right? yeah Exactly, exactly. But it is like extremely error prone and they make a lot of mistakes. And if you give them the same technology we're already using here in Western countries,
00:21:27
Speaker
It will save many, many thousand people, human lives, maybe 100,000 if it is applied broadly, per year. It's like yeah really large impacts that you can reach with this technology. not like It's not like a having fun with chat chat with you or I don't know, my girlfriend is like really about human lives.
00:21:48
Speaker
yeah so and ah But there is no business case in these regions and it is why startups and and and ah big companies are are not interested to you know to and enroll their technology there.
00:22:07
Speaker
at and It is why we started this project where we said, okay, we try to create an open source alternative for it. For sure it will be not so good as um as a of big companies or startups, sort funding but it can already create a large impact and also good for us in this country. The regulation is very low usually, yeah so it is relatively easy to to deploy and enroll such solution if we can prove that it works more or less okay. um And it generates added value for the hospitals. um And um you know I think such so it's projects are extremely important because again, so basically the same technology. yeah It is good for us in Western world, but we have already more or less functioning.
00:23:00
Speaker
quite good functioning, I will say, as ah healthcare care systems. Yeah. um I mean, we have an infrastructure we can build upon, right? Exactly, exactly. And as you know, if they're using this technology in our context, it is good, but the impact is, let's say, X. Yeah. And if you bring the same same technology into the developing world,
00:23:23
Speaker
the impact might be a hundred times six or even thousand more ah times six. yeah right And it is why AI for good is so important in my opinion.
00:23:36
Speaker
Now let's stay with this example for a second because I think it's quite telling. and we we We had a lot of debates about this. um Well, debates is the wrong word. I mean, we had a lot of discussions on how things could be done also with data and the training of the data and how we could, um how how it could be made available to to the people that potentially could use it. um
00:24:03
Speaker
Now specifically, when it comes to MedTech, that's basically what it is. What were some of the, let me say, the hurdles that you had to go through that you also didn't expect to, that we put, that we placed in your way when it came to this project?
00:24:21
Speaker
So um as i the in fact, we started with this project as we all were really small. It was me and maybe another of two guys. yeah and um ah And I was also working on it directly. And ah my first hard role where I was really dis encouraged. It was the availability of data, even data set are claimed to be open data.
Challenges in Open Data and Regulations
00:24:50
Speaker
So ah yeah, so in fact, so like I would say currently you have like 600,000 or something like this, maybe even already more um images of X-ray images of chest region ah labeled.
00:25:09
Speaker
that are claimed to be open sourced, but majority of them are available only for medical researchers. So I think it is like only 130K or something like this that is available for everyone basically. And they are really poorly labeled, but it is better as nothing.
00:25:29
Speaker
ah and ah but But the majority of open data is not actually open yeah and ah we could not get access to it because we are not a medical research organization. um The good thing is that um ah guys from another university, from one ah university, is they said, okay, let's kind try to solve this problem. They basically, they had access to every image available and they used it to create, ah to pre-train several models.
00:26:05
Speaker
ah as They said, okay, it's not for funau for commercial use or not for operations, but they make it open source and we build a upon these models. yeah So um improving them and and also building infrastructure around it so that it can be.
00:26:22
Speaker
used by hospitals. yeah They did not get into trouble for this? No. ah or instinct yeah It is like they open sourcing the model, not the images. They're like tricks people are using. Oh my God. This is Germany for you, I guess. yeah yeah and um Yes, it was like a great solution because it basically rescued the project because especially at that time, we were not be able to collect hundreds of ah sort of hundreds of thousands of X-ray images on our own. Labeled.
00:27:06
Speaker
label let label goods yeah yeah and so it's Just for the listeners to understand like label, label stands for um on this X-ray you see nothing and on this other X-ray you see something and also the region where you see it, what it is right there.
00:27:24
Speaker
A doctor looked at it and said there is something, has been checked by others. That's the data set the AI needs, which it's getting very expensive to produce such things. That's why nobody wants to publish their data.
00:27:37
Speaker
Yeah, yeah. Sorry for the interruption. Yeah, no problem. But it is in fact, it is like ah the next point is like the data availability is always a big question mark. um ah We are usually building upon some research so that we have at least some data. Yeah, but even in you know, like people will publish papers, say they have created something really great, but often data is not available or you must write to this person's, ah to the authors of the paper and they never, why? Of course not. Yeah. so So not working at the Institute anymore. And that email is not really working anymore. Exactly. Yeah. So it is often difficult. And yeah, but I think so currently, I have a feeling that the situation improves. Yeah, slowly, but it improves. And there are more and more ah
00:28:33
Speaker
um open data, truly open data. And also, especially in Germany, is there are some efforts ah from the governmental side ah to publish at least like um the let's say public, good, relevant data and also gather it from different sources, from different non-profits and, you know, like kind of organize them, structures them somehow and make it available to people and data scientists. Yeah. Mainly. Yeah. yes So I hope that this problem will be solved just in part soon, but it is definitely still a struggle to to get access to the data that in fact exists. Yeah. but
00:29:20
Speaker
But it is accessible just not accessible. And in fact, so for example, what also will be quite cool, um ah we never had the situation, but it is like, you know, like ideal world in my mind is for example, you you are a startup that gathered quite expensive data and created a cool application. Let's say met up medical application and in context of Europe or or US s and to sell it for big money and earn money with this, you know,
00:29:51
Speaker
give us free license for your software or give us your free data with an agreement that we do not publish it. ah So you're not free data. Yeah, your private data, but we promise to not publish it. And, you know, we built something similar for only developing world. And we sign a contract that we go only for developing world. And we go, no, we we never will compete with you. Yeah. But, you know, with the data you have, the algorithms you have, it is like,
00:30:22
Speaker
you know, people was talking a lot, that is new gold, that is new oil and stuff like this. ah soon Sometimes it is a it is Really, ah it's not a gold, or all it is ah like a wizard's step to solve so many problems of the world. And you know yeah not using it is almost a crime. you know yeah but but But it is not not obvious though for the most of the people. And we talked less about it. Not used capacities we basically have.
00:31:02
Speaker
yeah
00:31:05
Speaker
that will not even require to spend money or something. Just share it yeah ah in ah ah in in the ways that do not heart your business goals. yeah It's definitely possible, but currently we're not using it. And it is sad. That is sad. do you think Do you think we will get there? I mean, in terms of... to it to give like ah like ah Say a graspable counterexample, right? You also have pharma companies that invested millions, if not billions in drug development. And some of these drugs are more expensive in the West than they are in the developing countries. But in the developing countries, they still have access to it. Yeah, this is okay.
00:31:49
Speaker
you they can um Maybe at some point, maybe when the AI technology will be cheaper or more accessible or maybe if the market here in Europe and the US will be saturated, people will go to to to other another ah markets and if they do so, it will be great because their solution will be probably better as our ah solutions created by volunteers and it will be definitely a good thing.
00:32:18
Speaker
But still then, people are still dying. yeah so and yeah so In fact, I hope said ah um that either this scenario will happen, as you said, or we ah you know everything what we are doing is open source. and In fact we really wanted local experts because you know even in in in poor countries there' are still people who who knows a lot about AI and data science and are experts in it and maybe they can start with
00:32:51
Speaker
you know build upon our open source um applications and and create local businesses, local startups that you know understand ah the needs of the local market much better. um And can can react to it much faster. Yeah, it can react to it much faster and and deliver better services. That might cost something, yeah but is there better and still accessible?
00:33:16
Speaker
ah not too expensive, then hospitals will use them ah preferably compared to our solution and it will be definitely fine for me. yeah um Maybe this scenario, the second scenario will be also quite nice, but till we reach the situation, we must do a lot of non-profit work.
00:33:38
Speaker
Right. I mean, it's always I always say don't don't wait for things to get better. It will just take too long. I mean, you just got to do something now, if you can, right? That's the important bit. And whatever you can do at the moment doesn't have to be perfect, just has to move the needle a little bit. and say um Then can get the ball rolling. And just like with the um but the detection of ocean litter in very unexpected ways, you just could You just couldn't foresee at the beginning. Yeah, yeah.
00:34:11
Speaker
But and okay, so now we talked a bit about the problem that you face when it came to data. Now, when you had the open source model, were there also other hurdles? I mean, for example, legal things. I mean, youre you you are a nonprofit organization, but you're still a German nonprofit organization. Are there rules you have to adhere to even though what you provide is not even even provided in the European Union or even even not not even for sale? Yes, is as as there are many points. so ah What is definitely impacting us all the time is data privacy, GDPR, things.
00:34:54
Speaker
yes so It is ah because we are a German organization, we um must follow these rules regardless where we provide our services. And it is, in fact, sometimes really funny because, you know, we start to talk with hospitals, let's say in Ghana or Democratic Republic Congo, and you say, ah we need signs this and that, and there is this, you must ensure us that this and that happened with regard to data privacy and to look at us and be like,
00:35:26
Speaker
Who are you guys? What are you talking about? People are dying. yeah How can you talk about this yeah so i mean this? As a person living in Germany, I really enjoy this data privacy stuff. I feel better if my data are are protected. but yes you know we should ah When we especially talk about international international um In fact of it, we should keep in mind that data privacy is a wealthy world problem.
00:36:03
Speaker
yeah the Nobody bothers about data privacy in context where they are faced with um ah big problem, big crisis. In the worst case, if they their life is in danger.
00:36:20
Speaker
um And sometimes it is like really, really ridiculous in the situations we are in. We're discussing this with our partners. Also, of course, in the medical context, our medical project, we would never be able to publish them in Germany because know like medical standards so are so extremely high and we put so much money and efforts into various certifications.
00:36:46
Speaker
ah But since we are not applying this in Europe, ah we only give access it to people in the developing world. It's currently Africa is targeted. Then the situation in Africa is very different. say One should say also Africa is big. It's like very many many countries. and Every country has different regulations, but they are in general like easier as ours. In many countries, it is even like regulated like the hospital decides itself.
00:37:26
Speaker
what to use and what not to use. So there is no additional um check or enforcement. check yeah or enforcement Okay, this is usually it's like kind of certified hospitals, so they get a general improvement from the government, then they can decide what they do not use. And it is therefore, it is easier for us to bring our solutions to them.
00:37:50
Speaker
um ah But we will never be able to to you know bring it into the European market. yeah ah But it is also not our target. yeah yeah But in which way is then the the the data privacy thing an issue? Just give me an example. if data privacy ah yeah ah oh For example, you know we want to collect some some data on ah So this xwa project yeah um we um this interesting that there is It was proven in several the research papers that this public data on x-rays or AI trained on this data is biased.
00:38:34
Speaker
ah You can expect it towards a white male guy. yeah yes usually Just because more data points here. And we need more, interjet especially a small number of points, data points comes from ah from non-wide populations. yeah And if you want to apply this ah this in um and Africa, of course, it will work anyway, but but um can you can try to improve it. yeah And to improve it, you must gather examples from local populations. And we want we started to do so, ah but for example, ah in ah in the process of
00:39:18
Speaker
um It is, in fact, an example from Democratic liberal Republic probablyably Congo. In the process of working with images, the local doctors usually write names of the passions on the image directly. And it is something that they must do, otherwise the processes will not work. I see. Yeah.
00:39:43
Speaker
Or it will be, for them, why the difficult change in the processes must take into account. yeah and If we want we ask them for the images, they say, hey, you can have our images. And we say, oh, you know there is a name there.
00:39:59
Speaker
and when what yeah so that's a problem What's the problem? and yeah know but but Okay, data privacy. I really don't think that somebody here in Germany who have access to this data when they train this model have any intention to approach to some woman in some ah village and um oh ah Democratic rep Republic Congo and try to, you know, like I don't know, who do something bad to her based on the findings she can make or she can make ah on the 6-ray image. It is absolutely impossible situation. ah yeah But we are not allowed to have names on it. So we must find some some ways like
00:40:54
Speaker
It's so like it's so everything is manageable, but it like makes our life harder. Cost us time, of course.
00:41:05
Speaker
the down okay but But, you know, this is like, yeah, this is one example. There was another example where, you know, providing some, not so much AI, but more like data analysis for one Vietnamese organization that gives micro-credit credits for free to poorest families.
00:41:27
Speaker
so that they can build small businesses or increase their farming activities without interest. And we made some data analysis for them to analyze the efficiency of their campaigns. It was a relatively small project, but quite a cool one.
00:41:54
Speaker
you know um we had like sometimes spent to explain to them that we are not allowed to see names of people, of course, their addresses, the long takers. yeah And on the one side, it ah um It leads to the situation that we were not allowed to use paper documents. They have quite a lot of Word documents or paper documents with photos where quite interesting information is included, but always there is a name. And you know, there was no option for us to work with this data.
00:42:32
Speaker
without having access to the name. yeah And therefore we started, we said, okay, let us focus only on the structured data. They have also some access database. It was in fact accessed. And ze you knows they deleted the names.
00:42:47
Speaker
yeah but you know it this um this constraint like reduce the power of our analysis into special and consequently power of our advices to um to this nonprofit organization. And on the other side, side we just spend many hours on talking together. and on our side with lawyers on their side explaining them to why we need it at all. Then we send them like 20 pages document to sign something and something legal and says, guys, o I need to sign something but I completely do not understand why I'm doing this. So it's like the typical reaction. Yeah. So yeah, of course. And you must explain it. You know, you lose time. It is possible as the
00:43:37
Speaker
And again, here in this case, I cannot imagine that somebody on our site who was working with this data would come to the idea to somehow use ah ah the data of some poor farmer ah in Mecklenfeld to try to ah you know sell him something. or so sell them something but yes how it should through it work, yeah but but we must do this. But it is again, so, I mean, in general, I think data privacy is extremely important and GPR is in general good, but it is like we often see it only in context of
00:44:24
Speaker
big corporations, ah data crackers, I don't know, Google, Facebook, whatever, yeah, where you have fear of misusing our information, or maybe governmental organizations misusing information. And they're in fact, such danger here. But yes, but if you know, you introduce regulations that valid it for everything, you, you must also keep in mind that it might create damages too. And I will say in case of AI for good,
00:44:53
Speaker
Even such a TPR is like a bit older. I think they were not aware at all about AI in general, it's especially AI for good. But um this regulation makes life harder where it is not necessary. And in the worst case, you know,
00:45:13
Speaker
this losing Losing this time means ah bringing the working application ah later to the market, meaning that few people in the worst case died because they didn't have access to this. When the GDPR came out, i I mean, you know me, I'm insanely paranoid when it comes to data, way more than, I mean, people really call me paranoid. and um Maybe for good reason, maybe not, but it's fine. I'm um i'm putting these rules on myself to factor authentication, like all of this stuff. um But I do that for myself right because I know what can be done with this stuff. um I also can see why...
00:46:02
Speaker
I can see what the intention of the GDPR was. you know When it came out and I read through it, I first had to laugh a little bit because I thought, really, that's it. that that I've been speaking to my paranoia, right? yeah Yeah, really, that's it. And then, of course, what's going to happen, you see just two.
00:46:22
Speaker
I know this is not the limit of the GDPR, but what do you see on every website? website You see a cookie notice where everything except all cookies is already pre-selected anyways. Nobody gives a shit and just clicks on, yes, now let me watch this website or do whatever I want. Nobody actually reads any of this stuff or maybe ever says no anyways. So it's some parts of the GDPR.
00:46:49
Speaker
to me felt like, hey, let's force these companies to do that. The companies all freaked out because they didn't really know how to handle it because not every company ah has an IT data privacy expert on board anyways. And then what they did was just find a loophole to shift it to the consumer and the consumer has no idea anyways. Yeah. I mean, it just like.
00:47:18
Speaker
Often I have a feeling that many of regulations, and I hope it will be not the case with the AUI Act, but I see the dangers that it will be, is that that you know these regulations are made for lawyers or ah legal departments.
00:47:34
Speaker
It is basically of, you know, like finding a way of formulation to shift the responsibility to somebody else. yeah it's always It's always responsibility shifting, right? Yeah, it is exactly like the cookies. Yeah, you know, the cookie stuff. ah I mean, okay, or did read data privacy agreement, who reads this stuff? It is written by lawyers or lawyers that another lawyers cannot sue your company. Yes, exactly.
00:48:02
Speaker
At the end, so for me as a kind of consumer, it is like basically meaningless. I mean, what I would, associate maybe in this case, more... um It's going to cost you more because somebody has to pay for the lawyers, right? Yeah, exactly. exactly yes so ah but um ah I hope it will not be the case with the UI, but there is a danger that it will also happen, and then we will just have like...
00:48:32
Speaker
legal fighting, and every startup will have to employ three lawyers to, you know, just so not to change what they are doing, but to justify what they are doing. Exactly. It's always about justification, right? its Yeah, yeah. Except on anything else. Yeah, it it is then this will be, of course, not so so great s thing.
00:48:56
Speaker
ah But I still believe that the regulations are important in our And also, you act I believe it is a good thing that it happens. yeah I'm not very happy with how it was developed, but so in general, I would say it is better is nothing also for businesses and consumers and because you have at least some you know ah some frames in which you which you know you are moving. yeah so I think the intention is definitely good, but you know what they say, the road to hell is paved with good intentions. So we just have to be really clear that the conversation happens from the industry as well, from many perspectives, not only from
00:49:43
Speaker
The big tech corporations who can basically just send an army of lobbyists and lawyers and demand whatever they want. Because on the governmental committees, you don't often have the tech experts sitting there. yeah yes they They feel like they got better stuff to do.
00:50:04
Speaker
Yeah, exactly. and does this In fact, so in general, ah you know, here's a funny story about you act in some to some extent in my four people is coupled to you act because I don really I got to know my co founder manas um um after his talk about the UI act and he pointed out the problem that the creation of the act, um and it was like, 2020, so some years ago, this it was like, they just started to speak about it. And they were like, inviting, like, lubist, some maybe some ethical experts, maybe some
00:50:51
Speaker
ah Some researchers yeah to discuss about AI. Tons of lawyers, definitely. but no AI practitioners at all. So basically no one. So there were no people involved who know how to develop this stuff and how it works ah and what is realistic and what is not realistic. And his talk was in fact about, ah yeah or we he basically, um you know, like,
00:51:23
Speaker
um ah made a call for practitioners to know like ah get together and start to exchange and maybe start to create a small practitioner's lobby, trying to also influence um ah the act discussions. But um yeah, in fact, we we get in touch with this idea, but then we will just think it more about AI in general and kind of came to the idea, okay, AI for nonprofits will be also great, start to research on it, get research, identify the problems, and basically develop the concept of a machine. But you know, like here, for example, it was a UI act and I think, at least in in the early stages, but I think it's also in later stages,
00:52:11
Speaker
um ah the voice of the people who are actually working with data and who create added value out of it and who create applications, working applications, their voice was to,
00:52:24
Speaker
cool um not so loud as it should be, let's say. Yeah, and then there was a couple of organizations who were like, was associations of startups trying also to to be in to the discussion ah ah But in Zen, it was ah mainly big corpse and and and like lawyers who probably do not understand AI. On the level, it's huge to be able to create good regulations.
00:52:56
Speaker
yeah yeah I think we're going to talk about the the regulations. we We will get back to it, I promise you. but first i mean Thank you for already for for giving a bit more insight into the projects, right i mean the ocean litter detection, the radiology type of projects.
00:53:16
Speaker
but If we move now a step back a bit more towards, again, MI4 people, I mean, how do you, ah I mean, from a broader perspective, how do you approach these projects or how do you approach these projects in general?
Project Management and Nonprofit Funding Solutions
00:53:31
Speaker
I mean, what I mean is um the label nonprofit can cause many people to think very differently what about the product you produce, but also about the structure and the mode of operation of the company.
00:53:45
Speaker
So I would like them to get a better idea of how you do things and and how that might be different from maybe for-profits or how it's maybe not different. um One specific thing is you you've mentioned that you work with a lot of volunteers. I mean, that's so that's already um very different for for-profits or for-profits might call those interns, ah yeah unpaid interns. But you see what I mean? it's i i'm I'm trying to understand.
00:54:14
Speaker
um i I want the people to ah get a better feel of what technologies do you use? I mean, sure, it's AI, but is that different from the corporate people? Is there a specific type of resource allocation that you have to juggle? Which, I mean, for profits, well, what's your resource allocation? Get more funding and then just throw money at the problem.
00:54:36
Speaker
What about the hardware do you have to use with specific types of hardware or or how how does all that go and which of them gives you maybe even sleepless nights? I don't hope any of these, but I'm sure this causes a different type of stress than yeah and running a normal startup.
00:54:55
Speaker
Yeah, definitely. It's true. And it's like, I think I can talk about this topic like ours. ah But let me try to, you know, like comprehend it a little bit. um um ah Yeah, so maybe the first thing is like, organizational differences.
00:55:15
Speaker
i would say let's they was Let's start with similarities. Technology or tech stack in general is the same as and in commercial world. So um we use the same tools, the same techniques, also because you know like our volunteers, they are used to use it in their commercial, in their all day lives as employees. yeah And it is just easier for us.
00:55:42
Speaker
um What is the biggest difference is like how we um organize ourselves. yeah um Because we have only a few employees. yeah And um the idea behind MyForPeople or in our organizational structure is that Every project has ah one dedicated project lead. One project lead has several projects that he or she oversees. And these people are very experienced AI specialists who also understand not only how algorithms work, but also at least a little bit of cloud, a little bit of
00:56:28
Speaker
of engineering, a little bit of UI, you know, just they they can understand every aspect of end-to-end development of AI-driven application. um And these people have usually five, ten years experience, yeah? ah And so it is like senior guys.
00:56:49
Speaker
and um um and they ah they are employed and they are working full time on our air for good projects and they lead several projects at the same time between five and ten depends on on um ah um complexity of the project but the capacity is like largely five to ten for the person working in full time um and um um They are not like ah typical project managers. They are more like we call them technical project managers because say basically they plan everything what should be done um on technical side um and ah define out of it like working packages that are tangible and doable within relatively short amount of time.
00:57:46
Speaker
I see. and they give and then And then we have a a network of volunteers, quite a big pool of volunteers. And we pick those volunteers with the right skills for this particular project and this particular phase.
00:58:02
Speaker
Yeah. And then people start work on it. They get one working packages already with this they can use the next one takes the next one. At some point, either says we don't need their skills anymore, or they say if we have no time anymore, but you know, we have the progress that is like, how do you call it?
00:58:27
Speaker
Sorry. No problem. Like out. Incremental process, yeah. ah So that, you know, even if one volunteer says, okay, he or she has no time, we can easily find another one who can build upon creative work with relatively short onboarding time.
00:58:48
Speaker
um And through this process, we managed to run quite complex um and um challenging projects that involves a lot of different skills and and the um complex AI systems.
00:59:11
Speaker
So um this is basically on the organizational side. So maybe also how we get our projects or how we define them or how we choose them. It is also different as in let's say commercial world, ah nonprofits comes to us basically with their ideas.
00:59:29
Speaker
ah is it Is it always nonprofits that come to you or is it sometimes for profits that don't want to pay anything? In fact, it is ah um it is a little bit of a legal question. okay and It is in our charter, and it might be very special for Germany, I'm not sure, maybe in other countries too, but in Germany, if you so our um charter says that we develop AI ah research and on and develop AI for public good.
01:00:02
Speaker
In this public good, there are two different terms for it in German. One is called, ah comineful what you will basically translate as public good. yeah Or the my nu zika what is used interchangeably ah for public good and being tax-free.
01:00:27
Speaker
And in our charter, we had to write that we support this ah text-deducible organizations, because we wanted to write it more broadly, like public good organization, but then we will not get our nonprofit status. So it means that we are allowed to help only um um organization with approved text detection. Yeah. So, um and it is otherwise we we will not be able to be a nonprofit on our own.
01:01:01
Speaker
I see. Oh, I didn't know that. there Therefore, it is like, as in fact, sometimes I would help, I would like to help some, you know, startups that are maybe profit oriented, but they have cool ideas that has a lot of impact, and but they have no, ah no developers to do so. My opinion, it will be fine to support them, at least in the beginning, for the first prototype, um or in the people force, they can earn money, ah but ah but it we are not allowed to do so.
01:01:30
Speaker
um Yes, and does this is nonprofits come to us with their ideas. We discuss with them how how much sense does it make to use AI to solve this particular problem because often you don't need AI. Currently, a lot of people just believe AI is some kind of magic toolbox.
01:01:54
Speaker
that can solve everything, but not always yeah meaningful to use AI.
01:02:02
Speaker
The problems they want the solution for are scalable enough and important enough, and we have free resources and we start projects together.
01:02:16
Speaker
And to work very closely with nonprofits. But we have a couple of exceptions. For example, this marine litter project, in this case, we just said, okay, it is so important problem and it affects so much organizations and so many countries. So that we just start from right ah right from the beginning. We consider this project extremely global.
01:02:37
Speaker
We don't want to be binded to some regions or governments or whatever. We just do it on our own, talking to organizations that will use it later, like Greenpeace, WWF, Sea Shepherd, Freeza Sea, and so on. um But we are not biding this project to one organization.
01:02:57
Speaker
yeah um Yes, and it is basically how we operate. Again, technical stack is um the same. What is as in commercial setup, ah what is extremely challenging for us is to have ah to get funding.
01:03:13
Speaker
um Yes. um For for four many reasons. ah First, it is difficult to collect donations for AI for good because the topic is not very well known or the majority of people never heard about the idea of using AI for good. um It is something that we try to work on with our marketing volunteers, but it is like it will be a hard and long road. Yeah. yeah Educating the public costs time and yes effort. Exactly. And and also some ah public funding, or governmental fundings, what we in fact get a view, but they are very restrictive and no for in from many points of view. So first, they are always project oriented.
01:04:06
Speaker
And what we are doing is not, so our added value or our unique setting point is not why we're doing AI for good projects. Our main core, there's a core of our of activities basically providing an infrastructure to enable AI specialists all over the world to work on AI for good projects. So we basically this infrastructure.
01:04:29
Speaker
There's this guys managing all the projects, and the methodology behind it is basically our or main service, to be honest. yeah yeah um And I believe this is exactly what the world needs, because the idea is scalable in terms of ah number of we have for good projects, you can create per unit of time.
01:04:55
Speaker
yeah um Better as like they know like single projects where people somehow come together and start some open source things that eventually die after two months because people don't have time anymore to work on it. Yeah, yeah exactly. um and ah um ah But there is no funding for such type of organizations. We will need institutional funding and not project funding.
01:05:19
Speaker
and ah so it is quite difficult to get money. Therefore, in fact, we decided to be um and to change our operational model what we are currently in and we are currently in in the phase of transformation where we say, okay, ah we need money and we don't want to be dependent on very few donations and on ah governmental ah grants that are not really fitting us. um We said, Okay, we have this ah very smart people working for us full time. ah Why not to sell them as consultants for the 50% of their of their working time into the businesses, the corporations consultants. um Because you know, this is resources, human resource, resource
01:06:14
Speaker
How do you call it? In air quotes? cia Yeah, yeah, yeah. But I understand what you mean. Yeah, yeah, yeah. And um they um are very valuable. um People are or businesses are ready to pay good money for them.
01:06:30
Speaker
And if they work like 50% in commercial projects, they earn enough money to be able to run into the other 50% of their time already have a good project. Then we become like self-sustaining and sustainable also in the long run.
01:06:48
Speaker
Yeah, and um um ah it is why we start right now also commercial activities, not because we want to get to earn money. In fact, we are non-profit. We're not allowed to do any to make any profit. So everything what we order, we must put into our good project. as Otherwise, we will get any anyway problems um with the this is a that attacks departments audience one department um yeah um ah But it is also not not our intention to be for profit in any way. We want to just have enough funding to to be able to work on our effort, good projects in a sustainable manner and and ideally also in scalable manner. So scalability idea for us is extremely important.
01:07:37
Speaker
And you know if you depend on on governmental funding or or also ah foundational funding, it is like, okay, you get money for one, two years, you never know what happened after of those two years. In the long run, you must have something that as it is sustainable. and Or what happens a lot to other nonprofit organization, and it is really sad that you know they start with something, it was a good idea, wanted to do something.
01:08:05
Speaker
Then they search for money and notice, OK, there is no money for the things they are doing. And they start to change what they're doing. For two years, they get a grant, do something different, or not exactly the same what they planned to do. After two years, they need a new grant. They look where is the money.
01:08:26
Speaker
and ah and change again what they're doing in favor of this grant to be able to run it all. And it is not what I personally want for my people. So we have this project and I don't want to stop them prematurely. I want to ah ah to bring them all to the face where they actually generate impact and not just on paper, but on the field and are improving or saving human lives. Yeah. Right. And I mean, when it comes to grants, I mean, Germany is a rich country. It's a very rich country. um What I often see is that you have all these programs and it said this much millions have been made available, but it actually, I actually never know where this money goes.
01:09:17
Speaker
ah Sometimes I actually even think it never goes anywhere because people that try to apply for it just never get it. um ah Do you have any, you know, for these governmental institutions um when it comes to these grants, do you have any wishes what they would change? I mean, except from just having it project based?
01:09:41
Speaker
but Yeah, so um there are several, of us really is the biggest thinking, in fact, is focus on projects. And it is a it is good if you work manually and locally on something. But if you start to think about technology, technology, the power of technology is always scalability, at least of IT technology. yeah If you create something scalable, you don't need projects, you need infrastructures.
01:10:05
Speaker
that enables projects. And this is something what I believe people still didn't understand. so i mean um the Sometimes I think some deciders and like regulatory bodies, if you will try to explain them internet for the first part time, because I never heard about it, they would they will give you money to you know create Facebook oh yeah but But to not set up the infrastructure that is required that Facebook is running it. so um
01:10:43
Speaker
i oh see yeah and and And another thing is like, I would wish that um um ah ah google ah ah this bodies providing grants involve
01:11:04
Speaker
um a bit more technical experts already during the design of their programs. Yeah. In fact, they involve already ah technical experts decide on whether grants or applicants are good or not good. Yeah. That is definitely good thing. But again, already in the design, um, um,
01:11:33
Speaker
face It is extremely important to have understanding of technique of the technology. yeah For example, what ah currently is going on in AI for good sector, so we are not the only one who tried to do AI for good, but compared to others, I think we are the only one who provides this infrastructure and the rest of the actors, they have more like projects, but they're partially really cool projects.
01:12:00
Speaker
and um And there is already a little bit of money is that government gives for it. um And what I currently noticed, we have right now, rack-based AI for good chatbots popping out as mushrooms after the rain.
01:12:22
Speaker
And so, and you think so first. So, I mean, reg to implement reg, it is not so difficult. Yeah, but it's fine. I mean, also wall hanging fruit might be really great. Yeah. Sure. Yeah. But then, so if you like, for example, um um ah there are people who are doing, who are creating reg for, you know,
01:12:45
Speaker
um Helping people to apply for social help. A there is another project who is creating correct based chatbots to help people to apply for the social help. C, D, E, E, F and so on. And then what you think, for example, and you know, does they get money for this.
01:13:05
Speaker
ah just took Sorry, to interject quickly. Can you tell people quickly what this framework is? just so they wreck the no name and yeah basic Basically, the idea here is not to develop AI. It is basically um ah you have a database of documents or videos or audio or whatever, and then you um you use AI, pre-trained AI, for example, shell GPT.
01:13:31
Speaker
to enable users to interact with this database. So basically find the questions the or the answers to questions they have in this database. I see. And not going deeper in the details, it is a really cool application. And it in fact, it is a ah like ah increasing the power of this large language models quite considerably um ah but it is something where you already have platforms for where you can you know configure it's like within an evening partially if you have enough data yeah um and you know um
01:14:12
Speaker
And maybe this is they these platforms are not GDPR compliant or something like this. it There are many reasons not to use them. But yeah lino like from the perspective of a governmental body give money that gives money for AI for good, you know I should basically understand that I don't need 100 rug-based projects I need one rack platform for AI for good applications and enable 100 people to configure their application within short amount of time. right Because it will cost much, much less money. yeah yes yeah And ah it will have a better ah scalability effect also in terms of usage. It's an accelerator basically. yeah yeah say
01:15:07
Speaker
and Let's try 50 things. Let's let's but that start with the foundation of the infrastructure to enable the rest. Yeah, exactly. exactly and so you know and um Of course, you can also follow the strategy you like, okay, you give like 100 ideas, money, and let's see who will survive. And then you will believe that maybe it will grow also for different use cases. But it doesn't work because of this project thinking because you know, after two years, I have no funding more, how should they scale to other topics? Yeah. um And
01:15:45
Speaker
there should be another approach considering this very special area of rack based AI application for ah ah for public goods sector. What is in fact quite a cool low hanging fruit that can already create a lot of impact but you must approach it wisely otherwise It will not work out or at least the potentials of of this technology will not be realized completely. And and is this is why I think when people design brands already, say
01:16:23
Speaker
they they must talk to technicians ah to understand these things. Because right now for them, okay, AI, it's just AI something, then it is cool AI project. So you basically can sell them a RAC project.
01:16:37
Speaker
and And they will be fascinated by it exactly in the same way as you would say you would create some medical imaginary AI. For them it is the same and it is definitely not the same in terms of effort, but also in terms of impact. yeah yeah And they must understand it. And they are not technicians, but then they should talk to technicians already in the design phase.
01:17:04
Speaker
Yeah, for sure. I mean, you mentioned that the technicians, that there are some technicians or experts at least in you know checking the proposals for validity and things like this. um you know When I worked in some companies,
01:17:21
Speaker
i I wrote some of the these applications, or at or let me say, and sometimes I was also given the project lead on some projects where funding was acquired through, of course, you know writing these applications. So basically I asked what's the project, they gave me the The project proposal, I read this and my first question was, who the fuck wrote this? is this is What is this? this i mean And how did you manage to get the money? This makes no sense. But I mean, hey, okay, I guess we got the money. And then I was tasked to you know make something work to show off so we can get the next trench of money. And I was just thinking, how how am I supposed to do this? This this is terrible from front to end.
01:18:08
Speaker
And of course I know in that meeting that I had to present in front of another expert, right? So I prepared a lot and we worked things out and all of those things. And then, I mean, I should have known better. Of course I prepared for a level that was insane. And then after just saying 10 sentences, they were like, wow, this is great. Congratulations. And I thought,
01:18:36
Speaker
This is, this is what na I mean. Great. Thank you. But okay. I was just dumbfounded, you know, it was like, what is the, you, you're not even asking any follow up. What is this? This is not a test. This is not a check. This is just a, yeah, we had to talk to them and they seemed to know what they're doing. Check. What the fuck is going on?
01:19:04
Speaker
so and And those were the technical experts already involved, not trying to shit on them. They are, of course, also in a system which is trying to, I guess, self-sustain itself for all the wrong reasons. But when it comes to the technical experts, you said they should already be in the involved in the design phase. Now, when it comes to European Union, AI regulation, for example, the EU Act, or you mentioned other um um nonprofit organizations that try to do AI for good. Are there some um governmental you know initiatives being put in place where they try to do these things where you can already see, guys, you're missing the technical experts yet again.
01:19:58
Speaker
Yeah, I think so the dourings, this very long phase of development of the UI, since it was like very often I had the feelings that people talking about it it or also criticizing it or talking for it, um often miss technical perspective. um They were not necessarily 100% wrong.
01:20:23
Speaker
often and what they're saying, but they miss the technical context which leads to misinterpretation or in worst case misusing what's their proposal. whatsoever Yeah, yeah know there's like the classical stuff like I think they changed it in the end, but like before it was signed for a while, you know, the definition of AI could be interpreted as any Excel table. Yeah, so... Oh my God. Yeah, I guess this dude, you know, okay, to define what the eye is in fact difficult, but definitely not an Excel table. But you're right, you're right. When I read the first draft, I read the definition and I thought, wait, this is every software ever made. Yeah. What is this? Who wrote this?
01:21:07
Speaker
Yeah, exactly. And um and I think, and so what might happen with UAI Act is maybe similar to GDPR where you have like a good intention and maybe for for the cases you are talking about the whole time, it might be even a good solution, but you have done other cases you didn't consider that might get problems. Yeah, with GDPR, we talked already about it, like for us, for data-driven nonprofit organization working also worldwide, it's like really difficult, but it works, but it makes our life harder, considerably harder. And this might happen also with the UI Act, and I think often people, you know,
01:22:04
Speaker
i I believe it is also extremely hard to make a regulation that is simple, but at the same time fitting everyone's needs. sure But I think what people are not aware of is that they often might create something that is not always for good. yeah I mean, again, um When we are talking about application of AI in developing countries, and there could be an extreme impact, but in the worst case, we will not be able to to develop relevant applications for them because we are hindered by UI.
01:22:46
Speaker
yeah yeah and Again, so I mean, what they try to protect it or also people who, you know, like um are ah voting for even harder regulations, ah ah they have definitely good attention and they have good use cases for it too. But, you know, like pushing on this use case in the front and saying, okay, it is like the most important one. And in the end, it is like more about safety of our jobs, safety of our decisions, maybe
01:23:19
Speaker
um um safety of our privacy and data, so we're talking to Germany. This is important, definitely. At the same time, when you reduce this kind of of of of measures, you might hinder applications that save people's lives and maybe thousands of them. Is it good? so How to evaluate it ethically? I don't know, to be honest.
01:23:42
Speaker
I don't know. Because it's open. I don't know. Yeah, but I understand where you all the IT t data related regulations come from, but it makes sense. But, but often you also, you know, with regulation, you also ah not only binding the ah solution space for businesses or or for bad actors, but also for those who want to create something good.
01:24:13
Speaker
Yeah. I only have a few more questions left for you and um let's see let's see how fast we can get through them. Okay. I mean, just shifting back to the topic of AI in general and you work with so many different people, you worked on so many different projects and different types of company structures.
01:24:37
Speaker
Do you have any words of wisdom for other nonprofits and maybe what about for profits?
01:24:49
Speaker
This is the thing. I would say create purpose. This is what I basically learned in my career, especially at MI for people. You know, we were not talking about it, but you know, like if we have a volunteer application position, we often get like cert applications or volunteering position.
01:25:12
Speaker
and like compare it to you know companies that says a lot you know yeah companies fighting for people and we are not fighting at all we have no advertisements no people paid for the fighting them it is like just write a post and get certain applications it's easy for us to find these people because we give them purpose it is definitely for for non-profits it is non-profits often give purpose but then i will say try to give those people per purpose whom skill you need. Because often there's a mismatch between purpose and what I can do with my skills.
01:25:59
Speaker
yeah um And for for profits, definitely, I would say the purpose is like what you should create for your people, especially in the market where your employees can run over to another to to your competitors anytime. Yeah, because there are so few really good people.
01:26:21
Speaker
um And there but as as they are very asked by the market. um and Yes, and it is, I think what I'm trying to do at the MyForPeople, I try to give the people the sense of purpose in what they're doing. I think that's a good answer. Yeah, it enables extremely. and People with purpose are much, much more productive as people without.
01:26:49
Speaker
Yeah, the drive is just different. I mean, there there is one. All right. Now when it comes to purpose, ah so this fits actually quite well. So now it's it's time to think big, really, really big. So if you had the chance to work on any type of project with MI for people and people and money were not an issue, what would that be and why?
01:27:17
Speaker
Whoa. That's good project. It's it's it's Christmas forever. ah so um
01:27:31
Speaker
yeah so In fact, I think it would be something in the area of improving farming in developing countries.
01:27:44
Speaker
by it means of AI to fight poverty and hunger, especially, or something like a really big AI health care platform for developing countries with whatever AI medical field available, is available at this point that we you know like put everything together and find the shortage of educated professionals in these countries because it is like I mean, there's a lot of different impact you can create. And we have like so much problems. We have wars, climate crisis, and so on. But in Zen, so I will probably not be able to prevent wars. And for environmental AI, you can generate a lot of cool insights. But in Zen, it is like the decision of people.
01:28:39
Speaker
something or not to do something. But with the healthcare situation in developing countries, I really found it very sad because the knowledge is here, the tools are all here, but just because of some structural things that cannot be changed by one person or so or by decision of one person, they have not access no access to it and people are dying and human lives are the most precious thing.
01:29:07
Speaker
I can imagine and right so you know um be able to would to create such a platform, it will be like really easy way to, to and in the end, even cheap way, even so, yeah they need quite a lot of money, but in the end, it is like per per person life, it is like a couple of euros, yeah, and you could the do it, it will be really,
01:29:37
Speaker
Um, yeah, something really, really great. I would, uh, um, i I can imagine for me. Yeah. So that's fine. Great. Let's say, yeah, that sounds great. Now, I mean, I doubt there's.
01:29:56
Speaker
There's any book someone can read to to learn how you do what you do, but you know are there some some resources, books or courses or or online resources that you can recommend about topics we might have covered today that you yourself found genuinely helpful and that that you think people who would find interesting that listen to this? so I mean, of course, besides the MI4People website, of course.
01:30:26
Speaker
Yeah, of course. Yeah. So, um in fact, ah there is a couple of, you know, resources debating about time for good, not technical resources. Definitely. It's like more general, like, seeing from by that one month shift on the project called three frames, they published quite a lot. um Interesting stuff. Civic coding network in Germany, published quite a lot of interesting articles or also create webinars, more like high level webinars usually, and so again, non-technical, but ah with different, with interesting examples, so far, I forgot. These webinars are free to access. access so
01:31:12
Speaker
um um ah and um But especially ah in terms of advertisement of my four people, of course, yeah also a couple of ideas. so we i like ah We had to like one season of our own vodcast, where we are talking about the Africa in German, unfortunately, but German listeners could also ah take a look at it. This is available on YouTube, Spotify, Amazon Music, and so on.
01:31:39
Speaker
um um Unfortunately, it was only one season. I hope we can start with the second season at some point, but currently, we focus on other tasks. I'll put it in the show notes for sure, so people can quickly easily find it, of course. Yeah, yeah yeah thank you. It's called Kaiif you mentioned, oops yeah also a simple name. And also an interesting thing, because they are mentioning books, and ah There are many, there are several people and within MI for people so who recently came to me with the idea to write a book about Ashwagut. So not only through through the through the glasses of MI for people, but in general, because, you know, in the meantime, we we became quite connected to various actors just in Germany and this field. And we are planning to to to write a book about Ashwagut within the next year.
01:32:38
Speaker
And um yeah, that's great. Yeah. Then maybe is it then I will suggest to read it. Yeah. Well, then, then we will talk again. Yeah. Cool. and So if people, if people want to. um
01:32:55
Speaker
want to reach out to you, i mean and even if if they're just interested or if they want to work with you, sure, they can find you an MI for people. um where Where else can they find you? On which social platforms can they find you? so so We are mainly active on LinkedIn because it is like the place where we can find oh all the volunteers.
01:33:17
Speaker
oh um You know, like, of course, these volunteers are all also definitely on Instagram or or TikTok. but the ah But, you know, on LinkedIn, we're talking about technical stuff much more on Instagram. and It's a different level of conversation. yeah Yeah, exactly. It is a more natural environment for own to speak with our volunteers. and Unless someone's going to do a dance, then then it's then it's there on the other platforms.
01:33:46
Speaker
Yeah. And of course, also in terms of our relationship to businesses, you know, as this corporate consulting board, we are doing or we have a couple of sponsorship programs, it is like um more natural for us to connect to businesses or business leaders on LinkedIn as many other platform. So I'm available on LinkedIn anytime you can just connect with me and write me a couple of words. I'm I'm interested to to exchange and to chat with anyone. and Plus, of course, email that you can find on our web page. I'm always open to discussions. I definitely will answer it. ah That's great. and Also, all of this goes into the show notes as well, so you can find it on the website and and everything. Paul, ah thank you so much for you know taking the time on
01:34:40
Speaker
that you carved it out from your busy schedule because I know how busy you are. And um and yeah, thank you very much. Yes, thank you as well for the opportunity. It was a lot of fun. Yeah, it was a lot of fun. And thank you, of course, also for your contribution to MI4 people in our AI ethics committee. Of course. greate I feel honored to ask me.
01:35:01
Speaker
yeah Your fitting goals are really great with your skills in it. yeah And you think make a great contribution. And thank you for your invitation. It was a lot of fun for me to be here. I'm sure it's not the last time we talk on the podcast. yeah ah So thanks again and well to everyone listening. um Have a great day.
01:35:21
Speaker
Hey everyone, just one more thing before you go. I hope you enjoyed the show and to stay up to date with future episodes and extra content, you can sign up to the blog and you'll get an email every Friday that provides some fun before you head off for the weekend. Don't worry, it'll be a short email where I share cool things that I have found or what I've been up to. If you want to receive that, just go to ajmal dot.com. A-D-J-M-A-L dot com. And you can sign up right there. I hope you enjoy. it