Episode 5: Germany is Getting Clever with AI

Is there a German version of the famous chatbot? In what areas does AI have the greatest promise? Can international companies in the field find success if they set up shop in Germany? And why are German researchers so important in teaching AI to have a sense of humor?

Spotify

Apple

Deezer

Amazon Music

Google Podcast

The cast

Dies ist ein eingebettetes Bild | © Azadeh Ghahghai

Azadeh Ghahghaie:

Azadeh Ghahghaie is the director at the SAP.iO Berlin, which is part of the global network of equity-free startup accelerators of software giant SAP. There she helps promising startups to integrate with SAP solutions and accelerate their entry into a curated, inclusive ecosystem whose offerings can be easily accessed and deployed.

Dies ist ein eingebettetes Bild | © Antonio Krüger

Prof. Antonio Krüger

Prof. Antonio Krüger is an internationally renowned expert in human-machine interaction and artificial intelligence. He’s the CEO and scientific director and head of the research area "Cognitive Assistance Systems" at the German Research Center for Artificial Intelligence (DFKI). He also holds the Globus Endowed Professorship for Computer Science at Saarland University and is head of the Ubiquitous Media Technology Lab and scientific director of the Innovative Retail Laboratory (IRL) at DFKI.

Dies ist ein eingebettetes Bild | © Kian Mirshahi

Kian Mirshahi

Kian Mirshahi is the founder and CEO of Mayday.ai, a young company devoted to innovative artificial intelligence approaches for disasters and risk intelligence. Mayday has gained recognition and accolades from many UN agencies, as well the European and the German Space Agencies. Mirshahi is an advocate of using the power of AI to manage the impacts of climate change impacts and create resilient communities around the world.

Transcript of this episode

Presenter:

Hello and welcome to “Into Germany.” The German business podcast brought to you by Germany Trade & Invest, GTAI, the German government’s international business promotion agency.  I’m Kelly O’Brian!

Our topic today is, so to speak, the new kid at the head of the class. We’re talking, of course, about “ChatGPT” – the AI chatbot is amazingly good at writing poetry, solving coding problems, building apps - and also in doing kids’ homework. Which is why schools in New York have banned GPT to prevent a flood of artificially composed essays.

ChatGPT was developed by OpenAI, and it’s CEO is Samuel Altman. He says:

Altman:

Some Robots will eat some Jobs, and some Robots will kill humans. You already seen that with self-driving cars. But if you look at the trend at all, I think it’s gonna be incredible positive for humanity.

Presenter:

So where’s Germany in all this? As Europe’s leading industrial nation, you’d think that Germany would be a huge market for all sorts of solutions using AI. To find out more we’ll talk to Antonio Krüger, who runs one of Germany’s best-known research centers on AI. But first of all, let’s ask ChatGPT itself: what is Artificial Intelligence?

ChatGPT:

Artificial Intelligence (AI) is the field of computer science that focuses on creating intelligent machines that can perform tasks that typically require human intelligence, such as learning, problem-solving, decision-making, and perception. AI systems can be trained to recognize patterns, make predictions, and adapt to new information, and can be applied in various domains, including healthcare, finance, manufacturing, and transportation.

Presenter:

That was a mouthful. So ChatGPT, can you add some humor? 

ChatGPT:

Let me give it a shot: Artificial Intelligence, or AI for short, is like having a bunch of robots running around in your computer. But instead of just following basic rules, these robots are super smart and can do things that normally only humans can do, like learn from experience, recognize objects in images, and beat you at your favorite video game. So basically, AI is like having a bunch of mini Einsteins working for you, except they don't need to take coffee breaks or complain about their coworkers.

Presenter:

Okay, now for some HUMAN intelligence, courtesy of Antonio Krüger. He’s both a professor of computer science at Saarland University in the Southwest of Germany and the CEO of the research center DFKI.

Krüger:

DFKI stands for German Research Center for Artificial Intelligence and is nonprofit research institution that is run as a public private partnership. That means our shareholders come both from the public hand and also from industry. So we have major digitization companies, Google, Microsoft alongside with manufacturing companies from Germany, for example, Bosch and ZF, the automotive suppliers. And basically our goal is our mission is to transfer foundations of AI basic research into applications.

Presenter:

So, Antonio, has ChatGPT won you over?

Krüger:

I have to say I was really astonished to see also the quality of the output, you know, of, for example, ChatGPT, which is based on GPT and which, of course, also has been modified extensively, you know, in respect to what the original language model could do, you know, So it's very much tuned towards current application. And this is interesting in a way, because it gives us a glimpse on what might be possible if such a system is trained and specialized, for a certain area and domain. On the other hand, we also see we see these effects of hallucination. You know, I've just recently read the first evaluations actually of the large foundation and model that's built into Bing, you know, the Microsoft search engine. And it was pretty disappointing actually, the results. So it was just kind of too much nonsense in the answers. Yeah. And that's why you need somehow actually to develop these models further to have actually safeguards around actually the core of the model. You know, so that you that you don't end up with kind of confidently basically presented misinformation, ja, this is a little bit the danger here.

Presenter:

ChatGPT was invented in Silicon Valley. Why not in Germany?

Krüger:

Germany is not so well known for the big things, you know, for disruptive technologies, but for continued innovation of products in niche markets. I also believe that the language models that we see and that are successful knowledge, you know, ChatGPT and so on, they need to be grounded somehow, you know, grounded in terms put to use in factories, put to use in robotics and so on to learn actually from the real world. You know, and this what I'm pretty convinced about this is this would take away much of this detachment that the models basically exhibit nowadays, you know, these kind of hallucinations, it's like somebody said this, you know, this large language model. It's a little bit like an alien that tries to study a culture, you know, and has only secondhand material.

Presenter:

And this is where Germany comes in?

Krüger:

Germany at least, with the robotics, of course, co-robotics, you know, a very interesting platform to do this kind of things because they interact with humans directly. They can start to make jokes and then register the human reaction to those jokes, you know, and this will hopefully lead to funnier jokes and the current language models generate. You know.

Presenter:

So, Germans will teach AI a better sense of humor? I’m not sure the world’s been exactly been waiting for that.

Krüger:

No jokes, maybe not. But appropriate behavior, you know. An effective behavior, a behavior that is kind of aligned with our work culture. This, I think, is very relevant and needed. And this cannot be learned from rule books and from the Internet, you know, from Internet data. This has to be learned through interaction at the workplace. You know, I think this is obvious, and I think Germany is in a good place actually, to make make some contributions there.

Presenter:

Okay, let’s shift gears. In Germany, the driving force behind AI is the manufacturing industry, correct?

Krüger:

Germany, of course, you know, has a very strong background in manufacturing. So. And surprisingly, maybe but this is this is something that Germany and also Europe is very good at is also in robotics, ja, a special type of robotics, not so much the industrial robots, but the collaborative robots. So the robots that are able to basically work together with humans without hurting them in the same workspace. And this subdomain of robotics, this is actually an area where Europe we have very strong also European manufacturers. This, for example, a Danish manufacturer, but also in Germany, there are companies that develop these so-called Cobots - collaboration robots. And this I think is, for example, a strong a strong area.

Presenter:

So where else do you see opportunities in Germany?

Krüger:

There are, of course, certain areas where one can where we expect big impact actually of AI techniques, you know, and one is the whole area of health. And there are actually a couple of jump start ups and also initiatives that that make use of the AI techniques in particular in vision, you know, to analyze x rays, for example, and and make suggestions for certain types of diagnostic actually results and therapy suggestions. That's one part. And the other, of course, is in all manufacturing processes. This is I think for Germany relevant you mentioned, for example, recycling is an example. Also the circular economy can especially since we have the manufacturers that builds these machines, you know, that do recycling for example, that sort of materials automatically and so on. There has been made a lot of progress actually through the use of AI techniques. And I expect also this area to grow in the next couple of years. Another area where AI in particular in Europe would be absolutely necessary is a is the whole transformation of the energy sector. You know, you might know, especially in Germany, but in particular now also with the war in Ukraine, you know, this has actually accelerated massively the investments into renewable energy to enable independence in the energy sector. But these energy networks that are required for renewable energies that are mostly decentralized, small producers.

Presenter:

Okay, then tell me, how do these new energy networks rely on AI?

Krüger:

So we will end up in a more intelligent network grid, often this is called Smart Grid that uses the AI actually to balance loads to connect actually the best producer to the current consumers. So this is I think also something where AI will make a big difference. And that's something that will happen actually in the next 5 to 10 years. But then you need also clever mechanisms that use AI, that make forecasts. Actually, you will have connected systems, you know, that rely on weather forecasts to be able to to understand the demands of the future and of the near-future and that are then able to switch the grid accordingly.

Presenter:

Thank you, Antonio!

So you just heard Antonio Krüger from DFKI mentioning weather forecasts. This is where AI can make a big difference, because it might help to improve Predictions.

We’ll discuss this in a second with the CEO of Mayday AI, a German company which is using AI to predict catastrophes:

Kian Mirshahi:

 We have trillions of brains working together with different inputs and sensors and are building patterns. That's becoming their reference and their intelligence that's building their consciousness.

Presenter:

But now, as usual, let’s look at some other top business news from Germany

Fittest for Survival

Germans – like many others around the world – tend to replace their smart phones quite frequently, but the trend is toward longer-life devices. That’s according to industry organization Bitkom. In return, users say they are willing to invest more to get greater longevity. Curiously, average German consumers say they’re willing to pay  238 euros for a new phone while the average price of new devices is 563 euros.

New Record – Number 1

A public-private early-stage investment fund has raised some EUR 500 million in its fourth and latest round. That’s the largest sum ever in the 18-year history of the High-Tech Gründerfonds or HTGF. It currently has 1.4 billion euros at its disposal for German-based start-ups, particularly in the tech sector.

New Record – Number 2

Germany continued to set new heights in expenditures for research and development. The Federal Statistical Office says that 112.6 billion euros – more than ever before – were spent on R&D in 2021. That was an annual increase of well over 5 percent. The figure includes both publically and privately funded projects.

New Record – Number 3

Researchers at the solar-cell technology institute Helmholtz Zentrum Berlin also celebrated a new milestone. The scientists achieved an unprecedented 32.5 percent rate for converting solar radiation into energy using tandem solar systems. Tandem solar cells split the spectrum of light into segments and use a solar cell that is optimized to each section.

And finally, staying with renewables, Getting to the Grid

A start-up called CarbonFreed in the north of Germany has come up with an artificial intelligence application it says will dramatically reduce the time renewable energy producers need to get permission to feed electricity into the grid. The bureaucracy-busting system taking care of pesky paperwork, called “gridcert,” takes care of pesky paperwork, has already been used in 560 projects, the majority of which have been successfully completed.

Presenter:

Well how did that happen? We’re back at AI, and let’s go now to the university town of Darmstadt near Frankfurt. It’s a deeptech hub, with a strong academic focus on engineering and lots of big brains working on startups and university spin-offs. Kian Mirshahi decided this would be the best place for his business “Mayday AI”.

Hi Kian, tell us what Mayday does?

Mirshahi:

Mayday AI is a global company that provides real time disaster and risk intelligence, which which includes services around all stages of disasters. So before, when the risk happens, then detection of these events and then monitoring to post events activities. It is powered by the fusion engine, which is effectively capable of bringing in different data sources and automate intelligence around a certain event by effectively fusing these different data sources together. So so providing a 360 degree view and to to to give you some more context, for example, we're using a large constellation of geostationary and low orbiting satellites that we have direct access to. We have 35,000 traffic cameras, and we're also using social media sentiment analysis to to effectively bring intelligence around any events of interest.

Presenter:

Mayday can predict earthquakes? Could it have helped to warn people in Turkey and Syria?

Mirshahi:

So, you know, earthquakes of most disasters are probably some of the most challenging to date and we're seeing a lot of improvements because we simply don't have the data to be able to predict. We haven't been able to we have seismic activity. We know where the fault lines are. But there are things like gravitational pulls and pushes that are tied to other other planets, for example, the moon or other planets that are causing these big events that we just don't know. We still don't understand these behaviors. So there's a lot that we don't know. But an area that we have data and we have access to data we've been able to very successfully predict. So, for example, with wildfires, we can predict down to a property level for months ahead, you know, where wildfire could potentially happen because wildfires are predominantly a cascading effect of drought conditions. And we know exactly what drought conditions are happening. The same thing can happen can apply to flooding or other disasters that we're just starting to decipher by leveraging the power of AI to see more patterns that are conducive to these events happening.
 

Presenter:

How does Mayday use AI?

Mirshahi:

So, for example, if you're detecting fires on a satellite imagery, that's a two by two resolution from 29 from actually a lot more, 32,000 miles above sea level, the naked eye cannot see that. But you can decipher those detections and automations with artificial intelligence over time by training it just like a child. You know, a two year old toddler that goes through maturity and goes through that same maturity. Our brains build these models and mature artificial intelligence, which is our construct, is doing that in a manner that we haven't been accomplished as a species, in a collective manner, that we haven't been able to accomplish as a species. And consequently, it just makes us so much smarter. It will expand our intelligence and consciousness markedly by deciphering all these patterns and things that we just can't see, I couldn't see before.

Presenter:

You are also collecting data from social media – text and videos - to feed your platform. How does that work?

Mirshahi:

So we do audio to text convergence so we can effectively look for words and patterns in audio. So if someone says "wildfire" on a 911-call, we can pick that up. Or in Twitter, we look at proximity of a fire. if we detect it around that fire event, we gather all that intelligence on Twitter. So citizen reporters and folks on the ground who are shooting videos and bringing additional intelligence to firefighters who are not there. That's what we're providing. And at the same time, it could trigger an event. So if satellites and other sources are not working and somebody somewhere is standing in a place and they post something, we look for patterns around that and we see a few other folks are posting it. And it's a legitimate concern then we pick that and generate an event out of it.

Presenter:

Who is using your services?

Mirshahi:

So we are effectively offering solutions in disaster and risk intelligence for providing monitoring hubs and early warning capabilities to government agencies, to first responders, as well as in the private sector. We're dealing with folks who are in the paper and pulp industry, farming, agriculture, the insurance industry, and many more effectively. We're opening this set of services to anybody whose business is affected by by climate and deals with the environment.

Presenter:

You have customers from all over the world – in the US, in South America, Africa, Canada. Why did you set up your business in Germany?

Mirshahi:

Maydays history actually starts from California. This was a company that started its R&D in 2018, and then through a whole host of acquisitions and partnerships, moved to Germany, where we decided to make Germany our headquarters. We find Germany to be extremely exciting for a number of reasons, in technology, in AI.

Presenter:

Like what?

Mirshahi:

We found a lot of opportunities to really showcase and prove our value by working with the European Space Agency and the German space Agency initially. That allowed Mayday to be incubated in a fashion that quite honestly, some of the other markets were not ready for it.

Presenter:

What do you mean by that? 

Mirshahi:

So what I really like about what the European Space Agency is done is create this incredible constellation of satellites and and they've basically provided us for free for subscription and that in turn, getting access to this, they're democratizing this data has provided an incredible amount of value. Many, many folks use the satellites for even, you know, you mentioned Turkey's recent earthquakes centered on one center to say that satellites are being used everywhere to do analysis of interferometry and other things. Proposed event that's that's free. This is the beautiful vision that, quite frankly, that ESA and others had that's generated a model for others to follow.

Presenter:

And what made you come to Darmstadt? It’s not the typical place to choose for international companies.

Mirshahi:

So because of the space industry, there's a lot of R&D and very specialized talent around space. We felt this was an area that that could bring a lot of talent, German talent, specialized talent, we deal with highly specialized colleagues.

Presenter:

Most of the giant players come from the US or China. Do you see any examples from Germany that could bring up the next big thing in AI?

Mirshahi:

There are many exciting microsatellites startups that are coming up out of Germany. Germany can easily pivot look SAP is a perfect playbook. And Germany has been very successful. Okay. If you think of SAP as a success story, I'd love to see hundreds more of SAPs coming out of Germany because Germany truly can create this kind of scale if it decides to shift its investment in its thinking in that direction.

Presenter:

This was Kian Mirshahi from Mayday AI. He ended by mentioning one of Germanys biggest companies –SAP. And when we talked to Antonio Krüger earlier, he also specifically referenced to the software giant.

So what’s the deal with SAP? Well, for starters, SAP has just received a $13 billion offer for one of its subsidiaries – the AI-driven specialist on Enterprise Feedback Management, Qualtrics. They also run a startup accelerator in Berlin where they breed fancy AI babies. Meet Azadeh Ghahghaie the director of the SAP iO Foundry in Berlin. 

Ghahghaie:

The startups that we work more or less, all of them at the backend they have now A.I. or AML, as the backbone of their technology. And needless to say that AI provides a startup with their capability also to be more efficient, make better decision making, make better predictions, and gain a clear competitive advantage in the market. And the startups that we work with are generally for startups. By leveraging AI technology, they can drive growth, scalability and success in today's digital economy.

AI is becoming an important component of a startup's innovation and their tech stack basically. And it's across from industries. So from retail and consumer industries to health care, finance, agribusiness, so basically across industries.

And just to bring my company's perspective into the game as well. At SAP, which is an enterprise software company, we also love to support our enterprise customers to immediately optimize aspects of their business. So for us, a AI is helping the enterprises run intelligently and basically provide them with built in models for real life use cases and deliver intelligent and end to end business processes.

Presenter:

Thanks for that, Azadeh. What areas of AI in Germany do you find particularly interesting??

Ghahghaie:

Sustainability is one of those topics and areas where Germany can have a significant role. And when it comes to sustainability, also AI can help drive change in so many ways. So I think the combination of using AI in the field of technology is a space that Germany can have a major role and have a lot to say. If we look at AI as a tool, it can basically enable efficient business processes, such as creating leaner, more transparent supply chain that at the end of the day can help reduce CO2 emissions. If we look at A.I. as a technology to enhance sustainability, solutions, for example, are recommended now. So, for example, like a recommerce solution to support the circular economy that suggests pricing for reselling used goods. Or if we look at A.I. as a sustainable solution itself in its own right and its own place, it can help reduce, for example, power consumption, or it can help identify and detect food waste. So A.I. as a tool, as a technology and on its own can be helpful in tackling different aspects of sustainability like efficiency, CO2 emission, circular economy, and also efficient consumption. I would say.

Presenter:

Is Berlin THE German hub for startups. Where do you find companies for the SAP iO Foundry?

Ghahghaie:

Our startup ecosystem is dispersed across Germany. We of course have different hubs that there is more concentration of the startups in those regions. But Berlin and Munich area are the two strongest startup hubs and we can find really cool sustainability, supply chain transparency focused startups in those regions.

Presenter:

Can you name some of the startups in your accelerator and tell us what they do?

Ghahghaie:

SPHERITY is one of the startups that we work very closely actually. Berlin based and has been working with the SAP in the past series and trying to support with the whole supply chain alongside the SAP supply chain, support our customers globally. There is another great a startup is called GenLots Swiss based that they use A.I. and machine learning for the supply chain and supports enterprises on the procurement initiation. Supporting companies with identifying how much raw material from which source needs to be bought so that enterprises can run on lower consumption rate and lower its CO2 emissions.

Presenter:

Azadeh, you have startups all over the world in your portfolio – in the US, in New Zealand, in South Africa. What would you say is special about the German startup-ecosystem?   

Ghahghaie:

I have to say Germany has a very unique positioning when it comes to the A.I. market, giving its different characteristic Germany's strong academic and research institutions. Germany is a strong industrial base like look at our strong, small and medium sized enterprises institution and all the automotive and industrial manufacturing companies that Germany hosts are also that life science. You see all the amazing life science companies that are German based or they have started in Germany. I mean world players. You cannot just ignore the fact that our scientific engineering legacy in Germany, alongside the industrial players, give Germany a very unique point and also which is very, very important on the AI topic and it needs a whole new session is Germany's focus on ethical A.I.. Germany has been at the forefront of efforts to promote responsible and ethical use of A.I. and not just sporadically be so open with the regulations. Of course, some may say it's imposed some restrictions on the businesses, but in the long term, I'm convinced that it's it supports businesses to scale in a responsible manner. And this has included the development of A.I. ethics guidelines and initiatives to promote transparency and accountability when it comes to A.I. development in Germany.

Presenter:

Well, one could see this as a disadvantage on a global scale: Europe cares, other continents don’t...

Ghahghaie:

When it comes to the usage of the data and the openness to use A.I. in Germany, in Germany and in Europe generally, given the GDPR, we are more compliant and we are kind of standing as a united front in the framework of the GDPR. But in the long term, also the American companies, the American startups and anywhere else, if they want to scale to the European market, they would still join us in this whole initiative and mission to have responsible A.I.. So as of now, it may seem that we are it's Germany and Europe, but to have a global scale business, everybody should join this effort, I would say.

Presenter:

And with this, we have come to the end of this episode. But before we go, we’d like to tell you a bit about HOW GERMANY WORKS.

You just heard Azadeh Ghahghaie mention something called the GDPR. Here’s how to decode that spoonful of alphabet soup. The European Union’s General Data Protection Regulation, which came into force in 2018, required all member states, to enact national legislation protecting people’s private information. This led to Germany’s Federal Data Protection Act- the BDSG. There are lots of strict rules and regs connected with these acronyms, but they also offer big advantages. Companies that develop AI and other solutions in Germany know that they will be valid just about anywhere in the world. And you can be confident that YOUR personal data is being protected. And that’s HOW GERMANY WORKS.

We’ve come to the end of this episode. Thanks for listening. And remember, Germany Trade & Invest can help you bring your business to Germany.

Get in touch at gtai.com.

We’re also keen on your opinions, suggestions and questions. Please leave a comment in your favorite podcast app or drop us a line. You’ll find all the details in our show notes.  

Til next month – “Auf Wiederhören” and remember: Germany means business.