Interview: Eugen Gross, CEO and Founder aiconix
Eugen Gross worked for over 25 years as a cameraman and producer for various TV productions. He studied media management at the Media School Hamburg and lectured at the Media Academy Hamburg. In 2018, he founded aiconix. The start-up offers B2B solutions that aim to improve productivity in the media and creative sectors by giving their customers easy and fast access to the latest AI innovations. Their platform, an AI one-stop-shop, offers a range of features including facial recognition, speech-to-text conversion, captioning of images and videos, as well as additional proprietary features. We spoke to Eugen Gross about his path into the tech industry, his fascination with AI and the biggest challenges for young start-ups, among other things.
What is your fascination with artificial intelligence?
You can realise ideas with artificial intelligence that were not possible before. I originally came from television and never expected to end up in the software industry. But if you understand the principle of artificial intelligence, then there are opportunities in every industry to try things that were completely unthinkable before. That's why I'm fascinated by the approach the Finnish government is taking: Give everyone a basic understanding of AI! Because only then will you be able to use AI to create added value to your company, to your industry and to your environment.
It's interesting that I'm so enthusiastic. In school, the class below me was the first year to have a computer course. I thought to myself, I'm a year late, but I'm not interested in all taht. The interest only came later, when I moved to Berlin. In my friend's environment, there were a lot of computer enthusiasts who were doing their doctorates at the TU. When I transferred a 1.4MB floppy disk to my neighbourhood in 8 hours with a 14.4MB modem, I thought I was the biggest hacker. The technical interest started then.
What challenges are you facing in the field of AI at the moment?
It's insanely difficult to get training material for machine learning in Germany. We are currently building our own face recognition model. Instead of showing the model a picture of a person 100 times and training it that this is person X until the machine then also recognises person X, it should not even know who the person is, but recognise them automatically based on biometric features.
My clients have the problem that they rarely work with international pop stars or politicians, but with regional personalities. And if you search for the mayors of Spandau, Pinneberg or Dachau, for example, Google usually doesn't recognise them. But a regional media house has umpteen such pictures in its archives and has to find them. This only works if there is a model that does not deal with the person, but says in general "This person looks like this, gets an ID and I can always find this ID".
We trained such a model and tried to get training material. However, a lot of training material that is available on the web can only be used for research and cannot be commercially evaluated. So this material was out of the question for us. We therefore approached other actors and asked if they could support us. Some photo agencies did the same. But those who actually take it upon themselves to promote AI in Germany have not supported us so far. We asked the press office of the Bundestag, for example. The Bundestag has thousands of labelled pictures. We asked if we could have pictures of Seehofer, Söder etc.. We would also anonymise them and only use them as training material, not personalised. It took a very long time, but finally the answer came: they were not sure if they were allowed to do that. It was not definitely forbidden, but not specifically allowed either.
This shows the typical attitude in Germany: we don't go forward and say "we'll do it, we'll try it out first". Instead, we are cautious, reserved and prefer not to do anything for fear of making the wrong decision.
A friend of mine who founded a start-up in Austin experienced something completely different: she went to the mayor, wanted all the anonymised data on local transport as training material and was completely perplexed that she simply got it. Of course, there are different data protection rules there and here, but there is also a more open attitude and willingness to experiment. Our biggest challenge as an AI start-up is actually to get hold of training material. There is too little material and, above all, too little willingness to simply support young companies with their content and data.
What are you currently working on at aiconix in the field of AI?
On the one hand, we are currently working on our platform, on which we combine the services of many providers and refine their data. The idea for the platform came around when I was looking for a solution for the reliable transcription of audio files. I realised that there are a lot of services on the market: Google, Microsoft, Amazon - dozens of providers. Each of them is good at certain things, but never at everything. One is good at English, the other more Russian. One application is strong in facial recognition, the other in speech recognition and translation. When we realised that we needed a platform that unites individual providers, we saw that the market needed that too! Our first customers told us they wanted to use AI, but didn't want to deal with dozens of providers. And that's how the platform came to be what it is now. A one-stop shop for AI applications for media content.
On the other hand, we are also working on our own models. It's important that we also differentiate ourselves from the big providers. That we have our own facial recognition, for example, which is independent of the GAFAs.
What makes the AI solution at aiconix so special?
What is special is that we are a one-stop shop. We open up the whole world of AI to a company in a single place. I always compare it to the hotel booking engine: the customer may not want to sign individual contracts with individual hotels, airlines, car rental companies etc.. Instead, he goes to a booking engine where he can book everything together in one place. This is exactly how our solution works: the client does not have to worry about which provider is the best. The customer gives us the order and the use case and tells us, for example, "I want the best" or "I want the cheapest". And we do it for him. The client also has a fail-safe. If something new comes along, he can switch over and doesn't have to worry about these things. We support him with our expertise and he only has one interface to us. That makes our solution special.
What do you want to reach with AI?
I want creatives to take care of what they do best: Create content. They shouldn't have to type something out or search for images. How many times have I sat at the editing suite looking for that one visual I shot three weeks ago? I knew it was on the server somewhere. But I couldn't just press control-F and look for that scene with the sunset or with person X. I couldn't do that. These simple tasks can be done wonderfully by an AI. We want people to actually care more about their job again and not about administration, sorting or searching.
How do small and medium-sized businesses benefit from your offer?
In the past, we only addressed media companies. But now every company is actually a media company. Our target market is all companies that own media content. Whether it's a hospital or a state parliament - media content has to be stored somewhere, they all have MAM/DAM systems in the background. So in business terms, our market has suddenly become much bigger.
Even in the SME sector, everyone uses media content and can use AI for that. The easiest way we can pin it down is with language. No matter what industry you work in: You spend a lot of time in meetings, you have to convert spoken words into text, you have conferences that need to be summarised. It almost always starts with language.
We built an app that can be used in Slack. You drag and drop an audio file into Slack and get the transcription and subtitles back. And you can do it for a few euros a month. You don't need technology, you don't need to involve the IT department. You simply book an app in Slack and you already have added value. And this app has AI in it.
That helps everybody: Whether it's the student who needs to summarise 100 interviews for his PhD thesis or the SME who wants to transcribe audio files. No matter whether a television production needs subtitles or a hospital for a patient video. Or think of journalists who record interviews with an iPhone: As soon as they arrive in the newsroom, the interview is transcribed. For so many, this is a great added value for pennies. It's not science fiction, but a tangible tool that helps a lot in many situations in life, automates intelligently and takes away a lot of work.
We also do transcriptions for conferences, which is especially helpful in the current Corona times. How many times have I sat in a conference for hours, got a video afterwards and couldn't just press control-F and search for a very specific sequence with a keyword.
Live transcripts are also possible. This means that subtitles are automatically added to a livestream. This does not always work 100 per cent error-free. But many of our customers say they are legally obliged to have accessible offers. Our offer is a great help for this, small errors are accepted. Live transcription is also possible with a slight delay. In our live editor, an person-editor can check the subtitles again and make corrections before the stream goes online.
What will the influence of artificial intelligence look like in ten years?
Currently, there is a lot of discussion and talk about AI. That won't be the case at all in ten years. AI will be everywhere and simply be part of everyday life. It will no longer be a hype topic, there will be no more headlines in the newspaper and no more AI commissioners. The topic will have arrived.
But we also have to be honest and say: simple jobs will disappear. Some people will really suffer as a result. Truck driver, for example, is no longer a job of the future if there are self-driving vehicles. Or let's look at the industry I come from: A unique feature film or a creative commercial - it will be hard to make that with AI. Because AI is always oriented towards what has already existed and what has worked. It asks itself how the thing was received, how it can be improved. To do something completely new, completely different - that's where the creative person will continue to be in charge. But not to produce news stories where images from 20 sources have to be cut together six, seven times a day and a text has to be written. I think that job will be done by AIs.
In the past, when I was a cameraman, we had a lot of time. Five of us went out and had three editing days. Today, everything is much more compressed. Today there are automated cameras that replace the cameraman. Some jobs will change or become obsolete. At a football match, there will still be 20 cameras, but no longer 20 cameramen. Each camera will be programmed: camera 1 will follow this player, camera 2 the other one, camera three will always have the total view.
We really have to think about these people and find solutions for them. But as always, there are optimists and pessimists. At the moment, I'm looking at the positive side of AI with aiconix - and that's extremely exciting for me.