fischer Highlights - Vanessa Gieske, Principal AI Product Manager

00:00:00: Willkommen zu Fischer Highlights, dem Podcast der Unternehmensgruppe Fischer.

00:00:18: Welcome to a new episode of Fischer Highlights.

00:00:21: Our guest today is Vanessa Gieske.

00:00:23: She is an excellent person to talk to within our company when it comes to AI projects.

00:00:28: I'll be discussing with her how artificial intelligence is used in our daily lives,

00:00:33: how valuable AI can be for our work, and also what challenges exist.

00:00:38: My name is Wolfgang Pot.

00:00:40: Welcome to Fischer Highlights, Ms. Gieske.

00:00:42: Thank you for having me to discuss this exciting topic.

00:00:45: You're very welcome.

00:00:46: Ms. Gieske, let's say I've never had anything to do with artificial intelligence before.

00:00:52: How would you explain how AI works?

00:00:54: Basically, artificial intelligence is a software product that's capable of recognizing patterns

00:00:59: from a huge amount of data and then making predictions and decisions based on that.

00:01:04: If you want to adapt that to a slightly more visual example,

00:01:08: you can actually see it quite well when children learn.

00:01:11: For example, they learn about the different animals that exist through sounds or pictures they see repeatedly.

00:01:17: And AI learns similarly.

00:01:19: Only it really needs to be millions of examples until the AI model is good enough to accurately reproduce that animal.

00:01:25: Or even to create new creative animals from the components of different animals.

00:01:29: And fundamentally, though, I would also say it's a bit of a matter of definition.

00:01:34: What exactly is intelligent?

00:01:36: Because the term actually goes back quite a long way.

00:01:38: And even in the 1950s, there were first programs, be it chess programs or chatbots,

00:01:44: that were already referred to as artificially intelligent.

00:01:47: And opinions probably differ a bit on what's truly intelligent.

00:01:50: But nowadays, we generally speak of artificial intelligence when the system is able to analyze data

00:01:55: and also create new things from it, be it text, images or even compositions.

00:02:00: That's when we talk about generative AI and AI is now also able to understand us.

00:02:05: In that case, we refer to large language models.

00:02:08: That sounds incredibly exciting.

00:02:10: But you said it very simply and clearly at the beginning.

00:02:13: You spoke of a software product.

00:02:15: Others, if you look at the discussions about artificial intelligence around the world,

00:02:20: also say that artificial intelligence can think for itself, make its own decisions.

00:02:26: Can artificial intelligence do that?

00:02:28: So that's a totally exciting question and I think preoccupies a lot of people.

00:02:33: Especially if you observe the headlines, it almost seems as if AI will soon seize world domination.

00:02:38: But the truth probably looks a bit more unspektacular.

00:02:41: Hopefully.

00:02:42: Yes, namely the simple answer is probably no.

00:02:44: However, it's actually the case now that the so-called neural networks, these trained AI models

00:02:50: have become so complex that they are hardly comprehensible to us humans anymore.

00:02:55: And because of that, the whole thing can indeed seem very intelligent.

00:02:59: But ultimately, it's still based on a mathematical decision-making process

00:03:03: in which various parameters or millions of parameters are optimized.

00:03:08: And thinking is probably characterized more by human consciousness.

00:03:12: Yes, you've certainly highlighted a good distinction there.

00:03:15: Ultimately, we also want artificial intelligence to be intelligent.

00:03:20: Otherwise, it wouldn't help us in our daily work or private lives.

00:03:24: That's why I also want to ask, where do we actually encounter AI in everyday life

00:03:29: without even noticing it?

00:03:31: Nowadays, AI actually accompanies us throughout the day

00:03:35: from route planning to suggestions for new movies and series that match my preferences

00:03:40: or product recommendations and voice assistance that support us every day.

00:03:44: So AI is actually already very strongly integrated,

00:03:47: even if we don't always perceive it and it filters an enormous amount of information for us.

00:03:53: At the same time, new things like images and videos can now be generated

00:03:57: and they're getting better and better.

00:03:59: Currently, it might still be relatively easy sometimes to recognize that something was generated by AI,

00:04:05: but the systems are improving day by day

00:04:07: and eventually it will likely be the case that we can hardly distinguish between them.

00:04:11: You mentioned a good example, route planning.

00:04:14: Sometimes you notice that the system might not be that intelligent yet.

00:04:18: Or on the other hand, with the weather,

00:04:21: weather forecasts are largely based on AI.

00:04:24: That sometimes doesn't quite work out the way humans wish either.

00:04:28: So in that respect, the system definitely still has room to learn.

00:04:32: But you also said, how creative can artificial intelligence be?

00:04:36: For example, writing music or painting, all of that is already possible today.

00:04:41: Exactly. That's actually no problem for AI anymore.

00:04:44: Artworks, music, photos, everything's there.

00:04:46: And I'd say the results are already very convincing.

00:04:49: Ultimately, these new artworks are created based on existing data and newly combined.

00:04:54: And there again, it would be a matter of definition.

00:04:56: Is that creative?

00:04:58: For me, human creativity perhaps also distinguishes itself a bit by the intention and the story behind.

00:05:04: For example, a painting which then makes a difference in the way it's produced.

00:05:08: And what's the difference between artificial intelligence and robots?

00:05:17: Is there one you can define?

00:05:19: I'd say you can separate them relatively simply.

00:05:21: AI is the software, essentially the brain, and robots are the hardware, essentially the body.

00:05:27: Both can actually exist separately from each other.

00:05:30: For example, AI can generate or improve text purely as software,

00:05:34: and robots can perform tasks through simple programming.

00:05:37: But the combination is, of course, very, very interesting in order to make robots even more intelligent

00:05:43: allow them to act more autonomously.

00:05:45: - Now, let's talk about the business side.

00:05:48: What are some typical tasks that artificial intelligence

00:05:51: can take over in a company?

00:05:53: - There are many good examples now

00:05:54: that are already working and in use,

00:05:56: especially in quality control,

00:05:58: detecting errors and identifying rejects.

00:06:01: That has actually been working relatively well

00:06:03: for several years, and of course,

00:06:04: it's getting even better

00:06:05: with the new possibilities and technologies.

00:06:08: I also mentioned a few things earlier,

00:06:10: document creation, analyzing texts, analyzing data.

00:06:14: In the supply chain, for instance,

00:06:16: you can forecast demand or provide personalized services,

00:06:20: especially in call centers.

00:06:21: The use of AI is becoming more widespread,

00:06:24: but also in the daily administration environment,

00:06:27: from writing emails to creating images and presentations,

00:06:30: everything's included.

00:06:31: So AI will indeed take over more and more parts there.

00:06:34: - Now, let's think a bit bigger.

00:06:36: Will AI change business models in the future

00:06:39: or even create new business models?

00:06:41: - Definitely.

00:06:42: You can see that quite well

00:06:43: in the evolving startup landscape right now,

00:06:46: and especially because coding is now accessible

00:06:48: to everyone with artificial intelligence,

00:06:50: it's relatively easy to develop an app or create new code.

00:06:54: Accordingly, there are fewer and fewer limits

00:06:56: for new ideas and their implementation,

00:06:59: and therefore it will definitely have a big impact.

00:07:02: - That sounds very exciting overall.

00:07:04: You've already spoken broadly

00:07:05: about the opportunities for companies.

00:07:08: That all sounds very exciting,

00:07:10: and you've already talked about the many opportunities

00:07:13: for companies through and with artificial intelligence.

00:07:18: Now, let's also look at the risk side.

00:07:20: Are there risks that companies should be aware of

00:07:22: when using AI?

00:07:24: - Yes, definitely.

00:07:25: Actually, a major risk doesn't necessarily come

00:07:27: from the model itself, but from the data.

00:07:30: This includes data protection.

00:07:31: You need to be cautious when providing data to AI models,

00:07:35: because sometimes it's not entirely transparent

00:07:37: how it's processed or stored,

00:07:39: but also data quality itself.

00:07:41: If there are certain biases in the data, for example,

00:07:44: wrong decisions can be made in the end.

00:07:47: That's when we talk about a bias.

00:07:49: And regarding transparency,

00:07:50: there are increasingly more guidelines like the EU AI Act,

00:07:54: which calls for a certain transparency

00:07:56: about how models or applications work.

00:07:58: And especially in society, also very current,

00:08:01: there are fake news and deep fakes.

00:08:03: We're talking about deceptively real videos

00:08:05: that suggest something that didn't happen,

00:08:07: or Elon Musk's AI model, XAI,

00:08:10: which generated anti-Semitic statements.

00:08:13: So this clearly shows us that while there's great potential

00:08:16: with artificial intelligence, there are also dangers.

00:08:19: - If we draw a big line under this now,

00:08:21: where would you categorize it for yourself?

00:08:23: Do you see more opportunities or more risks?

00:08:27: - I see many opportunities that one can

00:08:29: and should definitely utilize to avoid being left behind.

00:08:32: So speaking from a corporate perspective,

00:08:34: at the same time, one should also be aware of the dangers

00:08:37: and simply handle the technology responsibly.

00:08:40: (upbeat music)

00:08:43: - When I listen to you,

00:08:47: I notice a certain fascination and enthusiasm for this topic.

00:08:52: And that's all the more exciting

00:08:53: because you didn't start out in this IT environment

00:08:56: or in the field of artificial intelligence as a subject.

00:09:00: You're a lateral entrant, so to speak,

00:09:03: and then dedicated yourself to this complex topic.

00:09:06: What fascinates you so much about it?

00:09:08: - Exactly, so I've been dealing with it for a few years now,

00:09:11: although perhaps more in the direction

00:09:13: of classic machine learning at first,

00:09:15: but I actually started my bachelor's degree in 2017

00:09:19: as a classic industrial engineer.

00:09:21: But even then, programming and data analysis

00:09:24: were a large part of the curriculum,

00:09:26: and I got totally excited about it.

00:09:28: It just captivated me,

00:09:29: and accordingly, I quickly decided to specialize

00:09:32: in digital business and management for my masters

00:09:35: and also wrote my master's thesis

00:09:37: on the topic of knowledge management

00:09:38: with artificial intelligence,

00:09:40: which is also a very typical application

00:09:42: for this technology,

00:09:43: and it's super exciting how the field is changing there now.

00:09:46: And that's where my fascination comes from.

00:09:48: - Yes, one can certainly hear that.

00:09:50: That means, of course,

00:09:52: you also apply artificial intelligence in your private life.

00:09:55: That goes without saying, doesn't it?

00:09:57: - Exactly, probably for the things I like less

00:09:59: or aren't good at.

00:10:01: For example, thinking of recipes,

00:10:02: especially when you only have a few individual things

00:10:06: left in the fridge that might not necessarily fit together

00:10:08: with the typical recipes you cook.

00:10:11: I always find it super exciting

00:10:12: to just give it to an AI model and see what comes out.

00:10:15: And it's often very, very good, actually.

00:10:17: I can only recommend it.

00:10:18: - Yes, we have something in common there.

00:10:21: I keep noticing that too.

00:10:23: If you have two left hands in the kitchen,

00:10:25: artificial intelligence helps you incredibly well,

00:10:29: and you can conjure up wonderful things on the table

00:10:32: with just a few ingredients.

00:10:34: What was the last dish you cooked?

00:10:37: Can you still remember?

00:10:38: - It was a one-pot dish with rice, spinach,

00:10:40: and chicken breast.

00:10:41: It was very good.

00:10:42: - So this recipe will be shared somewhere later.

00:10:45: We'll see, maybe we'll announce it within the company.

00:10:47: Ms. Gieske, thank you very much

00:10:49: for this exciting conversation

00:10:51: and for taking us into the world of artificial intelligence.

00:10:54: It was very, very interesting to have this future technology

00:10:58: explained to us in more detail by you.

00:11:01: And thank you very much

00:11:02: for your visit to Fisher Highlights.

00:11:03: - Thank you for moderating.

00:11:04: It was an honor.

00:11:05: (upbeat music)