Home » Jonas Andrulis: AI revolution and Europe’s tech sovereignty

Jonas Andrulis: AI revolution and Europe’s tech sovereignty

by admin
Jonas Andrulis: AI revolution and Europe’s tech sovereignty

It’s not consumers who need AI made in Europe, but business and civil society, says the founder of the startup Aleph Alpha.

Jonas Andrulis is co-founder and CEO of the AI ​​startup in which the German industry is placing its hopes.

Manuel Schlüter

You left a top job as a manager at Apple to found an AI startup in Germany. Why?

We are in an industrial revolution, and with Aleph Alpha we are at the forefront. We have a mission, we care about the tech location and the sovereignty of Germany and Europe. Now we are in a situation that I would never have imagined when we started: We had the largest financing round ever in this area in Europe, our partners are the best companies in Germany and the world.

Bosch, SAP and others are investing half a billion euros, a record for an AI startup in Europe. But compared to Microsoft’s 10 billion for Open AI, it’s little. Can you still catch up with the big tech companies?

It’s all a question of perspective. Of course, these companies have the larger AI models because they can invest more money in chips for AI training. But we were faster than Open AI with AI that can process images and text. And at the moment we are the only ones with explainable results.

Jonas Andrulis and Aleph Alpha

“The whole of Europe should hope that Jonas Andrulis is successful,” was the headline in Handelsblatt when the tech entrepreneur managed to collect 500 million in investor money for Aleph Alpha. With around 70 employees, the startup develops AI language models similar to Chat-GPT and applications based on them for companies.

Jonas Andrulis has Lithuanian roots and was born in West Berlin in 1981. He already founded the AI ​​startup Pallas Ludens, which was bought by Apple in 2016.

How does this work?

Our model makes transparent where its answers come from and shows where there is conflicting data. I’ll give an example: I say Jan is the new star basketball player in the NBA, and his parents were hobbits from the Shire. And then I ask the model how tall Jan probably is. Then the model says: “I have two observations about this. His basketball career shows that he is very tall. The fact that his parents are hobbits speaks against it.”

Are such features what are now making AI companies compete with each other?

It’s about creating added value with AI. The industry is still in its early stages. Voice AI alone is not a business model. Consumers’ willingness to pay for access to models like Chat-GPT is low, too low. You can’t make any profits with this. We rely on corporate customers. They ask, how can I transform my company with concrete applications. They don’t care what kind of language model is behind it.

So far there has been a race for the best language model, GPT4 versus the others. Is quality now stagnating and the race shifting to other areas?

See also  This is the date on which it will pass close to Earth – El Financiero

We already have far fewer hallucinations (false statements from AI models, editor’s note) than we did a year ago. Further potential for improvement lies in building structures around language models, such as memory. But there will also be innovation in human-machine collaboration: What could that look like, apart from the prompting of chatbots?

You talked about the industrial revolution. Will computers and robots take our jobs?

Work as we know it now will change. Just as muscle power was replaced in the industrial revolution, the monotonous part of knowledge work is now being replaced by machines. This creates more scope for creativity and complexity. If you look at the demographics, the question will not be whether we have enough jobs for everyone, but rather, how we can remain operational with fewer people.

So thanks to AI assistants we will be doing the job of several people in the future?

Yes. You can think of it as having an infinite number of interns. They can pick out things, pre-formulate them, and rewrite them. In the future, people will no longer be able to do such detailed work. I see humans as coordinators who understand complexity, take responsibility and ensure that the AI ​​does the right thing.

Aleph Alpha’s mission includes technological sovereignty and Europe as an AI location. Why should consumers care where their AI comes from?

Consumers don’t need to care. Neither do our corporate customers. At Aleph Alpha, our postal address is not our unique selling point. Technological sovereignty is not about isolating ourselves locally, but rather that we as civil society can also determine for ourselves what the world created by AI looks like and that companies and governments can act transparently and self-determined when it comes to technology.

And to do that you have to do voice AI in your own country?

In the past you could say: Well, then there is no social network from Europe, no semiconductor industry in Europe. But this new technology is different because it has a decisive impact on our thinking. The general public interacts with AI. I find it problematic that tech companies decide what is the right idea and how controversial problems are presented. Such questions should be decided by civil society; at the moment, it is the technology providers who decide.

See also  Free to receive "Runbow" + the well-received flight simulation "Drone Racing League", EGS will be permanently saved after receiving | 4Gamers

And you had the feeling that if you didn’t address the problem, no one would?

Of course, an American company could do that too. But the big tech companies are more likely to rely on a cloud monopoly strategy. . .

This means that they want to use AI to make their cloud more attractive and thus earn money.

The cloud market is lucrative, with a profit margin of 50 percent before taxes. We couldn’t pursue such a strategy ourselves anyway because we don’t have the resources to offer a cloud. So we are developing open source technology that customers can install – and choose which language model they use and where they run it. This is in the interests of the customers. You have more added value and room for maneuver, for example to switch if another language AI gets better.

Open source means open source code – software that is available for everyone to see and use on the Internet. How does this become a business?

Our principle is: We share our knowledge and publish our research. But we don’t just offer everything for free. When it comes to open source, there are various models that allow you to be open and transparent and at the same time make money through intellectual property, for example premium functions or that only large companies have to pay for licenses. We also use such methods.

One of your investors is the foundation of Lidl founder Dieter Schwarz, which has now also donated a lot of money to ETH in order to help set up an AI location in his hometown of Heilbronn. Will you also use the infrastructure in Heilbronn in the future?

The Innovation Park Artificial Intelligence (Ipai) is being built there. We are just an hour away and are participating as a technology partner. I assume that a leading AI center in Europe will be created there in the next few years. Very good universities are represented with the ETH and the TU Munich.

In France, Mistral, another AI startup that relies on open source, was able to raise a lot of money. Is that your biggest competition?

See also  Anycubic on sale: 3D printers reduced by up to 18 percent

It remains to be seen what exactly Mistral will do. The company is still very young. First of all, I’m happy that there is another strong team in Europe. And I hope we’ll coordinate so that we don’t do exactly the same thing. The field is wide enough for this.

What is the overall state of the AI ​​startup scene in Europe?

What we have achieved with Aleph Alpha is unique for Europe. When I started, it was said that you can’t do this with German or European investors, they can only do e-commerce. You wouldn’t get any financing for high-tech here. Our success shows that many people recognize that now is a special moment and that it is necessary to act courageously. On the other hand, we have been losing out to the US in terms of innovation for decades, and this trend continues. Europe is sluggish, but there is also a sense of will to transform.

The EU has drawn up the world‘s first broad set of rules on AI – is the AI ​​Act slowing down development?

It could have been worse. Nevertheless, in Europe we are behind when it comes to AI implementation, but we are at the forefront when it comes to regulation. For us, these rules mean that a lot of our energy and our resources are no longer used for innovation, but for compliance with the regulations.

We read that you enjoyed tinkering with devices as a teenager.

Yes, as a teenager I soldered equipment for amateur radio and then started doing commercial software development when I was 16. I also sent messages using the BBS, the bulletin board system, which was the forerunner of the Internet; it didn’t have a graphical interface, just letters on a black background.

A lot has happened since then. Do you think so much will change in the next 25 or 30 years?

The development is actually going faster and faster. Five hundred years ago not that much changed in a lifetime. A blacksmith was simply a blacksmith; in his lifetime there were one or two new ideas. That’s completely gone today because of technology. This is the catalyst that accelerates all processes. We are the same mammal we were 10,000 years ago. That does involve risks. At the same time, rapid change is something we as humans are actually built for.

You may also like

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Privacy & Cookies Policy