An AI startup from Heidelberg is taking on the big tech giants from Silicon Valley. In an interview, Aleph Alpha founder Jonas Andrulis explains what distinguishes his software from ChatGPT, what is important in the fight for AI supremacy and why a defeat would be so dramatic.
ntv.de: The text robot ChatGPT from the American startup OpenAI has startled the tech industry. The company you founded, Aleph Alpha, offers a similar service with Lumi and Luminous. Why hasn't the industry trembled before your application?
Jonas Andrulis: Aleph Alpha has only been around since 2019. Our financing is impressive in a European comparison. Of course, we are still a long way from the 20 billion that Open AI collected. OpenAI also does phenomenal marketing that has always made waves.
How groundbreaking do you think ChatGPT is?
ChatGPT is surprising and accessible like no system before. But it is also the logical continuation of the progress made in recent years. OpenAI and Microsoft have built a huge dataset that contains a whole lot of world knowledge and pop culture references. This of course makes the software playful, flexible, valuable and so much fun.
What are the similarities and how do the programs differ?
OpenAI is aimed at end users with its software. We are currently concentrating on solutions for companies or public administration. We look at complex work steps that ChatGPT cannot solve at the moment. Last October, the "Lumi" citizen assistant was put into operation for the city of Heidelberg, which is based on Luminous and always backs up all statements with curated sources. The system is able to respond to very individual, non-preprogrammed questions. Setting up the technology in such a way that it can also be used by the administration was an active decision. That's not a matter of course. But it was clear to us: A sovereign European AI must also be made available to the administration and promote its digitization.
A recent performance comparison has shown that your language model performed just as well as the competition. What does this mean for Germany as an AI location?
Not only have we reached that level of performance, we've also been much more efficient at it. Compared to the competition, Luminous has about half as many parameters and is therefore twice as efficient at the same level of performance. Almost no one believed that we would be able to do that beforehand. We received a lot of positive feedback and are also very proud of the results. Of course, we only compare basic models here. It is a step in the right direction and can trigger important decision-making processes. But that alone will not ensure our technological sovereignty.
What makes you optimistic that Aleph Alpha can keep up with tech giants Google and Microsoft?
I remain optimistic because pessimists don't start companies. The past shows that we have already achieved more than we were given credit for. We didn't just build technology that's on par. We have also shaped our own innovations and shown that we are not only very good at keeping up, we can also set the tone in some areas. Of course, in the future we will have to prove ourselves against the $20 billion that Open AI will spend.
How is that supposed to work?
For this we need our partners. If we now succeed in establishing our technology in the European economy, we have a realistic chance of technological sovereignty, which perhaps many would not have thought possible.
In order to be really dangerous to the competition, your company needs capital. At the moment, however, money is not particularly easy for investors. Are you afraid that this will be fatal to you?
We have to keep stepping on the gas now, because Microsoft and Google are doing the same. The tech giants have recognized that it is now time to decide who will shape the future of AI and who can dominate value creation. Of course, there have been better times for raising money from investors. However, artificial intelligence is the topic of the hour. Investors have also recognized this.
How big is the commercial interest in your AI language model?
So far we have not focused on making sales. Instead, we advanced the technology and formed partnerships. The first multi-million contracts are currently being signed. If things go as planned, these agreements could even exceed our sales plans. Our technology has the potential to deliver eight-digit value propositions for our customers. We're working on some big things right now.
Who are your customers?
Unfortunately, we are not yet allowed to name some very exciting customers. But that will certainly change this year. Who we may name are the central office for information technology in the security area ZETIS and the IT system house of the Bundeswehr BWI and the media house IPPEN.
The mass of companies that are now pouncing on AI shows that the battle for technological leadership has long since begun. One thing is striking: advances from Germany and Europe are the exception. How can this be changed?
In Germany, many entrepreneurs often lack the courage to implement. We may have phenomenal universities, but there is still a certain risk aversion and high bureaucratic hurdles. This makes it difficult for companies and inhibits innovation. Against this headwind, every innovation project in Germany and Europe must first arrive.
You used to work in a leading position in AI research at Apple: Why are the USA and China so many steps ahead of us?
In the USA in particular, people are more willing to take risks and pursue paths that can also go wrong. However, important strategic decisions are never without risk. In Germany nobody wants to make a decision that ends up going wrong. But that is part of it. If I'm not willing to take this risk, I can't set the big, decisive course.
What dangers does this entail?
Some US tech giants are worth more than the entire European market. This is nothing new. What's new now: AI is not an isolated market, but an enabling technology that will deliver gigantic changes in competitiveness and value creation in every industry and for every company. That means if we lose the AI race, we don't just lose an industry, we lose everything. We lose all control over our value creation, at least all information-based value creation. And that worries me. Because AI will shape our future like no other technology.
Juliane Kipper spoke to Jonas Andrulis
