Artificial intelligence at eBay: How we use it.
Artificial intelligence (AI) has long since become part of our everyday lives. Our podcast episode with AI scientist Dr. Aljoscha Burchardt showed that. The renowned AI scientist explained the complex technologies of AI to us in an understandable way and showed how they influence purchasing decisions in online retail and can be used profitably.
In this episode, we take a look at the use of AI at eBay. Christian Bornmann, Product Manager AI (Artificial Intelligence) Transformation at eBay, explains why we rely on AI, in which areas it is already being used and what that means for commercial sellers.
He also talks with hosts Isabell and Lisa about why we at eBay are committed to the responsible use of AI and why chatbots could be the future of shopping.
We hope you enjoy listening and good luck trying out our AI tools!
**Important links**
- Artificial intelligence at eBay
- Our key principles for the responsible use of artificial intelligence
- eBay Community
- eBay for Business on Facebook
**Mentioned podcast episodes**
** Chapter markers **
- 00:00:00 to 00:02:16: Intro
- 00:02:16 to 00:05:54: AI at eBay: How it all started.
- 00:05:54 to 00:11:55: Why do we use AI at eBay and where?
- 00:11:55 to 00:19:43: Which AI tools are we already providing you with?
- 00:19:43 to 00:28:05: Which basic principles do we follow when using AI?
- 00:28:05 to 00:32:39: Is artificial intelligence revolutionizing shopping behavior and, with it, commerce?
- 00:32:39 to 00:34:43: What should I do as a seller to keep up with AI?
Comment below the post in the eBay Community:
👋 Hi dear listeners! We’re looking forward to your feedback on our podcast. Did you enjoy the episode with Christian? Are you already using our AI tools, or are you not convinced yet? Let us know in the comments here!
Isabell: Is artificial intelligence revolutionizing shopping behavior and, with it, commerce too? One thing is certain: artificial intelligence, or AI for short, is a real game changer for e-commerce. Artificial intelligence can recognize patterns, make predictions, and improve the online shopping experience through personalization. But now you’re probably asking yourselves: am I even prepared for that? Short answer: yes! With eBay as a marketplace and our AI-based solutions, you are already well positioned today and also well prepared for the future. Why that is, you’ll hear now.
Christian: So first of all, for everyone who may now be thinking, oh my God, there’s this huge AI wave coming and it’s going to knock me over, and I’m missing all of it and so on. AI is fast and, for sure, will change many areas. But I’d rather call it an evolution and not the big revolution where you wake up tomorrow morning and realize, oh look, eBay has suddenly become something completely different.
Lisa: Welcome to “Alles top! Gerne wieder!” - the eBay podcast about commerce and e-commerce. I’m Lisa Haak, and today I’m not hosting alone, but together with Isabell, who is also here in the studio. Today we’re talking for the second time on the podcast about the exciting topic of AI.
Isabell: Exactly! In the first episode on the topic of “artificial intelligence,” we discussed the basics of AI in e-commerce. For that, we invited Dr. Aljoscha Burchardt as an external expert. He explained what AI actually is, why it is relevant for you, dear listeners, and where it is already being used in everyday life. The episode is also linked again for you in the show notes.
Lisa: Today we’re focusing on AI at eBay. We’ll look at where AI is already being used on our marketplace and what new features you can look forward to. Our guest today is Christian Bornmann. He is our Product Manager AI Transformation. That title doesn’t exactly explain what he does, because he works on integrating AI into our marketing platform at eBay. The goal is to make marketing even more personalized, targeted, and effective. And with that: hello and a warm welcome, Christian!
Christian: Hello! I’m really happy to be here.
Isabell: Yes, we’re really happy you’re here today too. But before we dive deeper into the topic, Christian, I’d like to know from you personally: how did you experience the beginnings of AI?
Christian: Yes, it wasn’t that straightforward for me at all. I first really came into contact with it, I’d say about five years ago. But back then it was different. Back then, people talked more about “machine learning” and “data science” and less about “artificial intelligence.” So it was mainly about analyzing data and making predictions from it. And I’d say my first contact with AI was rather frustrating, because it was too mathematical for me. And the setup back then was more like this: on one side you had experts, “data scientist analysts.” They were amazing with data. And then they had to sit down with the business people to figure out how we could use the data to make our business better. But for me, that was rocket science. And I really tried to get into it. Between us, because no one else will hear it: I even considered studying a bit of “data science” on the side, but I gave up after the first semester because there was way, way too much math and it wasn’t for me. And we also always had projects, so as long as I can remember, we’ve had data-driven projects here somehow…
Isabell: By “here” you mean at eBay?
Christian: Here at eBay, exactly. But for me it was more like, well, we probably have to do this and I hope it works. And my revival of the topic came about a year or two ago. When “generative AI” suddenly came up, and then, as some of you have probably already tried: “ChatGPT” and the like. Where you can suddenly interact with it as a human using human language.
Isabell: Yes, in the episode with Aljoscha about AI, we also learned that AI actually originated more than 20 years ago in a way. But the boom, as you just said yourself, only really came in recent years. So when did eBay start dealing with AI?
Christian: Well, I think it really went exactly the same way for us. If you want to use AI to further develop your business, you basically need two things, in my view. One is a clear task. What still doesn’t work well anywhere is: “Oh dear AI, help me be more successful.” Nothing would come of that, of course. But a clear task, something like: “Should I phrase the text one way or another so it gets found as more interesting?” or “Should I send my email in the morning at ten or in the evening at eleven or something like that?” A clear question helps. The second thing you need is data. AI lives on looking at, in quotes, as large a data pool as possible, and then recognizing things in it that we as humans may no longer be able to see at all with our human perception. And eBay has both. Our task is crystal clear. What we are here for and what we want to do is completely clear and understandable, even for an AI, and we have data coming out of our ears.
Isabell: So that means AI has also been happening at eBay for five years now. So it’s not something that only started in the last year or two when it became a trend, but we were there relatively early?
Christian: I, in my area of responsibility, started using it commercially at eBay five years ago. But from my point of view, we’ve been there from the beginning. AI probably just used to be called something else. So, as I said, a few years ago I heard “machine learning” first. And people also resisted the term “AI,” meaning artificial intelligence, because there was nothing intelligent about it, it was a data pattern or something. So if you say that the different branches of artificial intelligence are included, I’d say eBay has been there from the beginning.
Lisa: Isabell said in the intro that AI is especially super relevant for e-commerce. If you break it down, why does eBay rely on the use of AI now?
Christian: As I said, we’ve basically been using AI for a long time. But until now it was still this data-driven “machine learning AI.” And since we now have these language models, which you hear so much about at the moment, “generative AI,” we suddenly have an AI quality that allows, on the one hand, talking to the system and, on the other hand, letting the system read things or look at things and then continue using the information obtained from that. And I’d say that’s already a quantum leap. And in that sense, the potential is then enormous, not just for us but of course for all other companies too.
Lisa: And Christian, if you haven’t dealt with the topic as much as you have: you keep mentioning “generative AI.” Can you explain again what that means?
Christian: “Generative AI” is a relatively new capability of artificial intelligence in the public sphere to interact in language constructs. So, for example, if you used to say with “machine learning”: “Okay, if we move the sending time of an email from 11:00 to 11:30, the click-through rate or whatever, the open rate, increases by so many percent.” Then that is of course important and interesting and so on. But it is difficult, very difficult, for normal people in everyday life to use. “Generative AI” means you can communicate with the system in normal human language, basically. So you can say, for example: “Dear AI, I want to write an email, and in the email, it should go to business customers and it should deal with a specific topic. And the tone should be this and that, and so on.” And then you let the AI run, and then it will come back with an email text, for example. And then you can say: “I still don’t really like it that much. Now I’d like it rewritten again, please. Please make it more ... highlight one point more than the other, or maybe make it shorter.” Sensational feature! You’ve got a text that’s 30 pages long, and you say: “Please, please, I don’t have time - no joke - summarize it for me. In two paragraphs, and pull out the top three points.” Snap! Five seconds later you’ve suddenly used “generative AI” to save yourself a lot of time.
Isabell: I think that interaction is what makes it special, that you can really interact with an output once more.
Christian: That’s sensational! I mean, I think the people who, for me personally but also in public, were the first to really bring it out. That was when, for me, at the beginning of last year for the species that were still on it earlier, already in October before that, “ChatGPT” suddenly came out from “OpenAI.” And they really set it up exactly like that too. So this language capability can be used in a thousand places, but the moment you build it into a dialogue, it becomes a completely new experience!
Lisa: And now let’s translate that into the things it actually gets used for at eBay. Your title says you work on the marketing platform, but beyond that, what initiatives or topics at eBay involve AI?
Christian: In general, I’d like to divide that first into three different areas that you can look at qualitatively. One is how AI can be used to optimize internal workflows and processes within eBay so that they simply become more powerful and more customer-friendly. These are things that people probably don’t necessarily notice from the outside right away. Things like using AI very nicely to write much more program code much faster. As a customer, I may only notice that indirectly. But if developers can suddenly produce or deliver new features much faster, then that is of course an advantage for me. Or customer service, where AI is used to support the employees on site, perhaps. So these are internal things that you as a customer don’t necessarily have to notice at all. The other is very, very concrete. The second category for me would be something like: how can we use AI to really improve the user experience of the user? And of course that works wonderfully on both sides, both on the buyer side and the seller side. In other words, what can AI do to make the shopping experience on the platform even more pleasant, even better? Or the selling experience? We may come back to that in a moment when we talk a bit about concrete examples. So that would be the second pillar. And from my point of view, the third part is again more invisible or only felt very indirectly, and that is the question of how AI can be used to basically achieve the actual purpose of the marketplace: bringing people who want to buy and people who want to sell together. Where can AI be used to make that even better? So to connect them even better, to help one find the other, for example.
Lisa: I’ve never understood it this well before as I do right now. Thanks, Christian.
Christian: Too kind!
Isabell: Okay, so AI improves the selling experience. What does that mean specifically for our sellers?
Christian: First of all, for everyone who may now be thinking, oh my God, there’s this huge AI wave and it’s going to knock me over, and I’m missing all of it and so on. AI is fast and, for sure, will change many areas. But I’d rather call it an evolution and not the big revolution where you wake up tomorrow morning and realize, oh look, eBay has suddenly become something completely different. And that’s how I’d understand our approach here too. We’re continuously looking at what developments there are, what technologies exist, what potential each one has. And then we try to build concrete, and at first also quite small, building blocks from that, AI building blocks, so to speak, which we then let flow into the product.
Isabell: And what building blocks are we building right now specifically? Is there anything already that sellers are benefiting from today in this evolution?
Christian: So, um, yes, of course! When you think about where AI can be integrated and where it can be used best, you usually look at areas that seem complicated, time-consuming, and difficult, and you check whether there’s anything where I can use AI to make this problem easier to solve with the possibilities we just discussed. And I’d also interpret our roadmap in this area in that direction. For example, a feature that is already available in Germany for all sellers is the ability to generate an item description with AI. Anyone can try it right away. You’ll find it in the app, for example, with this icon that has these typical little AI stars on it. If you click on it, it generates something. And the response we get to that is extremely positive, because it’s very simple, it saves time, it helps people, and in principle you can then create more listings much faster and ultimately do more business on eBay.
Isabell: Yes, we also often hear from sellers that they don’t use it to replace their own texts, so to speak, but to supplement them, because they’re nice texts too, so they use it more as a source of inspiration.
Lisa: It’s almost a bit like a synonym search sometimes. I used it privately and was totally amazed by how eloquent our AI is.
Christian: Yes, unfortunately there’s nothing to add to that, I’d see it the same way. And I think it’s a very nice example of how you start using AI without suddenly having completely changed everything. It just makes things more pleasant, better, and therefore very helpful.
Isabell: Are there any more tools we already offer?
Christian: It all goes in that direction. Another classic is: oh no, images. My image isn’t good enough, or I need a different image. That means a lot is happening there too. For example, how can AI be used to replace or remove the background from an image?
Lisa: Also super nice, especially when we think about the fact that a background, for example, must not even be there for third-party platforms in order for us to be able to advertise those items. If we give our sellers a tool that simply cleans up the backgrounds, I’d say that can also just be an increase in sales.
Christian: Absolutely, and above all it simply removes a barrier. Because if I’m a seller and I’ve got, say, 1,000 or however many items, listings on the platform. And then I see, oh yes, I need to remove the background. I know, okay, maybe I’ll do that the day after tomorrow or something. But with our friend, friend AI, you can possibly do something like that much faster, and that’s in everyone’s interest.
Isabell: Yes, and above all it’s then just one click and no longer this erasing by hand that we remember from the past.
Christian: That’s a nice example and you can also, of course, take it further. Something that’s being worked on, but isn’t yet available in Germany, is for example the idea of creating an eBay listing directly from a photo. And for me that is dreamland. I have something lying around or standing there that I think, yes, I could actually sell that now, but do I have the time right now? Then I just take a photo, and boom I have my listing in the system and can sell immediately without spending any more time on it.
Isabell: So the AI then automatically fills in the category and the item specifics and probably uses the AI-generated item description directly as well?
Christian: Exactly, that’s used too, exactly! And this feature itself is, I’d say, currently in some kind of beta phase. So it’s still being tested. And what it will look like in the end, very, very exactly, and where and when it will be available, also depends a bit on what other insights come out of it. But that’s exactly the direction it’s heading in. The idea is: I see something and want to sell it, and I don’t want to deal with the details of creating the listing, but rather have the AI do that for me.
Lisa: Super beneficial, then, for every seller listening right now and thinking about how many product managers or product managers might have to be consulted to find the right description or the right title, because you need to know the product well enough, that you may not need that anymore and can use resources elsewhere, because listing on the platform is simply so clever that you can then use the good internal resources more for product sourcing. Or something like that. Huge time saver!
Christian: Yes, and you can already see that experienced sellers will now say, yes, okay, but what if the AI sees something wrong and suddenly the wrong version of an iPhone is listed and goes out at the wrong price? And that is exactly why I also expressly say this is something we are currently looking at, experimenting with, and which is not yet available broadly. Because the devil is really in the details, and of course we absolutely do not want to make the process easier but then introduce errors. So there is still a bit of work to do there, for sure, but very promising and I think that direction is definitely worth pursuing.
Isabell: Super exciting! Now, we’re also using it in the ads area, can you tell us a bit more about what exactly is happening there right now?
Christian: Yes, that’s actually also a classic use case for AI. A very nice field for AI, because if you want to optimize ads, and if you don’t have a lot, a lot, a lot of “know-how” at that point, it can very quickly become very time-consuming. Because you have to look at the different data sources. You have to think carefully about the strategy and how you want to structure it on eBay. And for something like that, AI is really very well suited. Looking at large amounts of data and understanding what this ad is about, for example, and who might see it, or in which channel I want to place it. That’s a very good topic. So that’s already been running for a long time. The unfortunate thing for us here today is that this functionality is basically in the background; it falls into that third category I mentioned earlier. What can we do to improve it? But you don’t really see that much of it, unfortunately!
Lisa: But maybe sellers do feel it in the results.
Christian: I’m very sure they do.
Isabell: I was just about to say, the news isn’t actually that sad if I know that what I’m seeing is already basically AI-based.
Christian: That’s true, of course for me: then nobody even notices it, so you can’t really celebrate it. But yes, that’s true, the perfect AI would be something we wouldn’t even notice, it would just be in the background and simply make our lives nicer and better. But we’re probably still a few steps away from that.
Isabell: Where we can still see or notice AI or use it consciously is in the social media area. That’s going back toward “generative AI” too. What exactly is happening there right now, Christian?
Christian: Also a very nice field for AI, because everything is also a question of time. You had a listing, and then you want to promote it or post it on different channels. And then you’d actually always have to think carefully: wait, which channel is this now? And how do I do it now? And do I really have the time for that? So it’s probably more of a nuisance in many places, and maybe it often just doesn’t happen at all. And AI can help wonderfully there. Telling an AI: “Look here, this is my listing and I want to put it on this channel,” then the AI generates a very nice post for it. And then the seller just has to use it to post it.
Lisa: Christian, you talked about this wave that can kind of overwhelm or overtake you a bit. And AI is already a topic that makes many people uneasy or worried simply because they’re not so close to it. We stand for the responsible use of AI, but what does responsible use of AI actually mean?
Christian: Well, there were really two points in that. And I’d also like to say something briefly about the first one. Namely, for everyone who opens their email in the morning, Twitter, or whatever channel they use to get informed, and tries to avoid the topic of AI because every day it feels like they’re being flooded and oh God, I can’t read all of this and I don’t even want to. But actually I should, I can only say: first of all, I don’t know if it helps, but I feel exactly the same way. Even though I love dealing with it, it’s a lot of fun, I still often enough, I’d say several times a day, feel like, really? Already this fast? So don’t worry, that’s completely normal and completely human, and that shouldn’t stop us from dealing with it. What’s important, and I see it this way too, is that we are completely clear about how we use this technology. Because the possibilities are, I don’t want to say unlimited, because that would of course be nonsense. But there are a lot of scenarios in which AI can be used. But for us as eBay, it has always been extremely important from the very beginning that we use certain ethical principles when dealing with AI. And there are also, online, various sources. I think that’s very high quality, of course, but also as a private person it speaks to me strongly, and I’d like to mention my top three, just briefly here, and also pass them on to everyone who wants to use AI, no matter in what context. It’s really important to realize that there are risks and that you actually have to think beforehand about how to deal with them. It is not a good solution to take your customer database and throw it somewhere into AI to see, yes, what else could I improve there or something? Because once the data is in there, it’s in there. And to slowly return to your question, dear Lisa. My top three AI ethics points are: One, or the most important one for me, is to ensure that an AI is not biased. That sounds totally crazy now. Why should an AI be biased? It’s actually just a computer program or something, right? But you have to understand that an AI is nothing more than a certain algorithm that has processed, let’s say, unimaginably large amounts of data in order to learn from it. So, for example, if there are clouds in the sky, then it rains, no idea, just to take a very simple example. But this can go arbitrarily deep. And then there are truly awful examples. For example, you tell an AI: “Please generate an image - and I’ll make a few examples now - of a manager, doctor, prison inmate, model.” And if you look at that, what will come out, for example, you absolutely have to do this with Nobel Prize winners for a thing like that. And without commenting on that further, when ten images are generated, then with very high probability nine out of the ten images are old, bearded white men who then get a Nobel Prize.
Isabell: And how can I avoid that?
Christian: Only the person who essentially controls this process can moderate that; it’s called “training.” So this AI is, at the beginning, basically stupid, and then it gets training material. For example, images — it gets images of scientists, images of prisoners, images of this and that, and learns. And then it understands, very simplified, aha, that’s a typical prison inmate or a typical tree, a car, whatever. But if you just give it random images without control, then the AI will afterward simply reflect back what it found in the images. And there were, this was a few years ago now, but there were attempts by quite well-known “players” too. They went ahead and said: “Oh man, great, we have an AI, let’s switch it on and let it learn online.” They simply read news sources and followed tweets. And I think it took no more than three or four days before they shut the thing down again, because it was highly racist, misogynistic. It simply absorbed the mood of what it learned from. It’s basically like a child: if you take a child that is born completely neutral and it grows up in a society that is not open, transparent, fair, or diverse, then it’s very likely that the child will also develop corresponding values.
Isabell: Okay, so to tie it back to the principle you just mentioned: it’s important, so that the AI is unbiased, that the people feeding the AI information also essentially “monitor” that information and don’t let it run out of control.
Christian: Yes, I mean, that topic alone could fill an entire evening or even weeks, if you will. But in principle, yes, that’s right. And where that doesn’t work, I can’t, not everyone can train their own model, because that’s a huge amount of work too. But then those who use it have to make sure that it is not, so that this is taken into account when using the model. Super important, and there’s no excuse there either. Everyone who uses AI must first understand this problem and think about mechanisms for dealing with it. The best thing is to do the “training” properly. So that would be the first point. Super important: when we use AI, no matter where, whether I’m doing a product search at some point or in customer service or wherever. A person deserves to know whether they are interacting with a human or with a robot, computer, or AI right now. That means this must be a basic principle: when we use AI, we make very, very transparent where we use it, how, and what we access with it. Important basic principle. And the third point is that when dealing with AI, you always have to look at how we protect the user’s privacy. So you very quickly end up in a chat setup, not necessarily, but it really lends itself, as we discussed. And in such a chat, as the provider, we have only limited control over what is entered into that chat. And it’s important to make sure that nobody accidentally enters data there that shouldn’t be entered there at all. So that would also be a third important point.
Lisa: Super important to hear the basic principles for your own behavior, and it’s also very, very nice to see that eBay puts a lot of thought into this. Lisa: And if you want to hear more about these principles, of which Christian just listed three: they are public, and we’ll link them for you in the show notes as always. Coming up, we’ll return to our question from the intro: Is artificial intelligence revolutionizing shopping behavior and, with it, commerce? But first, we want to hear from you out there. Are you already using the AI tools to create item descriptions, item images, or social media posts?
Isabell: And if so, let us know what you think of the tools. If you’re listening to us in the Spotify app right now, you can also do that directly via the “Q&A function” in the app. Just drag the “slider” from the bottom edge of the app upward and off you go.
Lisa: If not, feel free to write to us in the eBay Community or on our Facebook page “eBay for Business Germany.” We also welcome your feedback on the podcast and, of course, a like or a subscription where you’re listening to us right now.
Isabell: So, Christian, now back to the question we’re all so eager to know: Is artificial intelligence now revolutionizing shopping behavior and, with it, commerce too?
Christian: You always have to be a little careful with the word “revolution.” Will commerce change because of AI? Absolutely! Will it happen overnight? I don’t think so. I think if I were to speculate a bit about where this could go, then I’d simply say that in one, two, or three years we’ll all have gotten used to surrounding ourselves with these kinds of “AI assistants.” That will be completely normal for us. It will simply be in contact with us in our messenger apps. One of them will be a fitness advisor for me. It can then say: “Hey, you need to go running again soon!” So for all kinds of life topics, that will be completely normal. And those will be AIs, and they’ll advise us like that, and logically there will also be AI assistants taking care of the topic of “shopping.” And I actually believe that user behavior will definitely change over time because of that. We’re used to searching via search engines - whichever one it is. And that means we enter something we know we need, and then we get search results. And then we decide whether we want to buy them or not. In an AI-driven environment, it will be more of a conversation. That means I’m now talking to a virtual assistant, an AI, and discussing with it what I actually need. And no longer: “I need a concert ticket for concert XY.” But rather something like: “My brother-in-law, whom I only see twice a year, who likes sailing and is turning 50. And I have no idea what to give him. Please!” And then the AI says, probably the AI will ask questions and say: “Yes, sailing is a good start, what kind is it? What are your favorite colors or whatever?” Then you’re suddenly in this dialogue, and in the end you’ll still, strictly speaking, get something like a search engine result, but much more tailored and with much smarter hits. And that will be, that will already be a new, that will already be a new shopping experience, I think. The other thing I’d expect is that classic product purchasing will increasingly merge with other activities. So, for example, today if you’re planning a birthday, you’re planning a birthday party. Then you have various activities, like picking music, inviting people, preparing a buffet, and however else you celebrate parties. And that always includes buying products too, for example. And today that’s traditionally separated. That means you do your party planning and then you pick the people and then you write an invitation and whatever else you do. Right, and at some point you also order some products. Fine! In the future it could be that you’re actually talking to your party bot and saying: “Yes, I need here a, I don’t know, birthday for a child of such and such an age or for me or for whoever.” And this bot then takes care holistically of what you need. So it writes the invitation card for you, creates the playlist for it, figures out which people you could invite, and so on. And I know, Lisa, you look horrified. Oh no, how can the AI possibly know all that? I know, that brings us to data privacy. But that means we’ll still have to see exactly how that will work in detail. But in principle, it will be like this: we could all have a kind of guided shopping experience where an AI takes us by the hand and helps us get exactly what we actually need.
Lisa: What you read on my face was more fascination, because I’d love to have more chatbots for all the topics in my life.
Isabell: And I find it funny that you’re using the birthday party example right now, because I actually just used AI to plan my birthday party. Especially with quantities, what could I make to eat, what’s there to drink? I invited X people and so on. Definitely super exciting if that ever gets combined with: “Buy it directly here!”
Lisa: We recently had it calculate how much “crushed ice” we’d need to buy for a party with 100 people, of whom how many are buying alcoholic and non-alcoholic drinks. So you’re not far off!
Christian: Wow! That’s very good, very nice!
Lisa: Christian, if I’ve listened carefully to this episode and I’m a seller on eBay myself, then I’m a little left with the question, “what’s next?” So leaving this episode, what do I do now? Do I wait until eBay comes to me with information? Do I keep myself informed? What should our listeners do now?
Christian: A bit of everything you just said. What I would absolutely recommend to everyone is to definitely keep track of what’s happening right now. Strictly speaking, no matter on what level. You don’t have to spend hours every day dealing with AI and what’s coming. But if something comes through the media, take a look and try it out yourself. So your example of planning parties or ice consumption or whatever, excellent! Really good, because you suddenly get a very personal feel for it and can then much better judge whether this is something for me in a certain situation or not. The big problem for sellers is of course that you already have plenty to do, and becoming an AI expert on the side is difficult, and I wouldn’t even recommend it, because it makes no sense if 100,000 people do that. And that’s the point, and maybe that’s also my main message here today for everyone listening: eBay, as a partner to commerce, is also a partner to commerce here. That means we really have a lot, a lot, a lot of attention, focus, and activity centered solely on looking at what’s happening right now, where it’s going, what’s useful, and what may not be useful at all. There are also things that pop up, get totally hyped, and then disappear again. Too bad for everyone who spent time on them, because it was basically wasted. We can wonderfully cushion that. So being on eBay as a seller is definitely a good ticket on this AI train, because we make sure that what is sensible and necessary and possible is made available in a way that is ethically clean, reasonably developed, and well managed.
Lisa: That was a very, very nice summary! And attentive listeners will know what comes now. Because, Christian, at the end we always ask: what was the last thing you bought on eBay?
Christian: Yes, I, hmm.
Lisa: Did an AI tell you what to buy?
Christian: That would be nice, that would be good! Then it wouldn’t be, I mean, lately I’ve bought, I checked again, there was a lot of toys and books involved. And that’s all ultra boring. But I need a new bike. That means the prediction of what I’m going to buy next is much more exciting than what I’ve bought, and that will be a bike.
Isabell: An e-bike or a regular bike?
Christian: No, I mean, I love e-bikes, but for health reasons it wouldn’t be an e-bike.
Lisa: Then let’s see what you’ll be coming to work on in the future. And with that, we’re already at the end of our episode. Thank you so, so much, Christian, for the exciting insights into the world of AI and for the many, very, very lighthearted explanations. It makes us super, super excited for the future. We’re looking forward to everything eBay will still provide there. Thank you so much, Christian!
Christian: Thank you very much, it was fun.
Isabell: Thank you from me too.
Lisa: Goodbye!
Christian: Bye, tschüss!
Christian: So first of all, for everyone who may now be thinking, oh my God, there’s this huge AI wave coming and it’s going to knock me over, and I’m missing all of it and so on. AI is fast and, for sure, will change many areas. But I’d rather call it an evolution and not the big revolution where you wake up tomorrow morning and realize, oh look, eBay has suddenly become something completely different.
Lisa: Welcome to “Alles top! Gerne wieder!” - the eBay podcast about commerce and e-commerce. I’m Lisa Haak, and today I’m not hosting alone, but together with Isabell, who is also here in the studio. Today we’re talking for the second time on the podcast about the exciting topic of AI.
Isabell: Exactly! In the first episode on the topic of “artificial intelligence,” we discussed the basics of AI in e-commerce. For that, we invited Dr. Aljoscha Burchardt as an external expert. He explained what AI actually is, why it is relevant for you, dear listeners, and where it is already being used in everyday life. The episode is also linked again for you in the show notes.
Lisa: Today we’re focusing on AI at eBay. We’ll look at where AI is already being used on our marketplace and what new features you can look forward to. Our guest today is Christian Bornmann. He is our Product Manager AI Transformation. That title doesn’t exactly explain what he does, because he works on integrating AI into our marketing platform at eBay. The goal is to make marketing even more personalized, targeted, and effective. And with that: hello and a warm welcome, Christian!
Christian: Hello! I’m really happy to be here.
Isabell: Yes, we’re really happy you’re here today too. But before we dive deeper into the topic, Christian, I’d like to know from you personally: how did you experience the beginnings of AI?
Christian: Yes, it wasn’t that straightforward for me at all. I first really came into contact with it, I’d say about five years ago. But back then it was different. Back then, people talked more about “machine learning” and “data science” and less about “artificial intelligence.” So it was mainly about analyzing data and making predictions from it. And I’d say my first contact with AI was rather frustrating, because it was too mathematical for me. And the setup back then was more like this: on one side you had experts, “data scientist analysts.” They were amazing with data. And then they had to sit down with the business people to figure out how we could use the data to make our business better. But for me, that was rocket science. And I really tried to get into it. Between us, because no one else will hear it: I even considered studying a bit of “data science” on the side, but I gave up after the first semester because there was way, way too much math and it wasn’t for me. And we also always had projects, so as long as I can remember, we’ve had data-driven projects here somehow…
Isabell: By “here” you mean at eBay?
Christian: Here at eBay, exactly. But for me it was more like, well, we probably have to do this and I hope it works. And my revival of the topic came about a year or two ago. When “generative AI” suddenly came up, and then, as some of you have probably already tried: “ChatGPT” and the like. Where you can suddenly interact with it as a human using human language.
Isabell: Yes, in the episode with Aljoscha about AI, we also learned that AI actually originated more than 20 years ago in a way. But the boom, as you just said yourself, only really came in recent years. So when did eBay start dealing with AI?
Christian: Well, I think it really went exactly the same way for us. If you want to use AI to further develop your business, you basically need two things, in my view. One is a clear task. What still doesn’t work well anywhere is: “Oh dear AI, help me be more successful.” Nothing would come of that, of course. But a clear task, something like: “Should I phrase the text one way or another so it gets found as more interesting?” or “Should I send my email in the morning at ten or in the evening at eleven or something like that?” A clear question helps. The second thing you need is data. AI lives on looking at, in quotes, as large a data pool as possible, and then recognizing things in it that we as humans may no longer be able to see at all with our human perception. And eBay has both. Our task is crystal clear. What we are here for and what we want to do is completely clear and understandable, even for an AI, and we have data coming out of our ears.
Isabell: So that means AI has also been happening at eBay for five years now. So it’s not something that only started in the last year or two when it became a trend, but we were there relatively early?
Christian: I, in my area of responsibility, started using it commercially at eBay five years ago. But from my point of view, we’ve been there from the beginning. AI probably just used to be called something else. So, as I said, a few years ago I heard “machine learning” first. And people also resisted the term “AI,” meaning artificial intelligence, because there was nothing intelligent about it, it was a data pattern or something. So if you say that the different branches of artificial intelligence are included, I’d say eBay has been there from the beginning.
Lisa: Isabell said in the intro that AI is especially super relevant for e-commerce. If you break it down, why does eBay rely on the use of AI now?
Christian: As I said, we’ve basically been using AI for a long time. But until now it was still this data-driven “machine learning AI.” And since we now have these language models, which you hear so much about at the moment, “generative AI,” we suddenly have an AI quality that allows, on the one hand, talking to the system and, on the other hand, letting the system read things or look at things and then continue using the information obtained from that. And I’d say that’s already a quantum leap. And in that sense, the potential is then enormous, not just for us but of course for all other companies too.
Lisa: And Christian, if you haven’t dealt with the topic as much as you have: you keep mentioning “generative AI.” Can you explain again what that means?
Christian: “Generative AI” is a relatively new capability of artificial intelligence in the public sphere to interact in language constructs. So, for example, if you used to say with “machine learning”: “Okay, if we move the sending time of an email from 11:00 to 11:30, the click-through rate or whatever, the open rate, increases by so many percent.” Then that is of course important and interesting and so on. But it is difficult, very difficult, for normal people in everyday life to use. “Generative AI” means you can communicate with the system in normal human language, basically. So you can say, for example: “Dear AI, I want to write an email, and in the email, it should go to business customers and it should deal with a specific topic. And the tone should be this and that, and so on.” And then you let the AI run, and then it will come back with an email text, for example. And then you can say: “I still don’t really like it that much. Now I’d like it rewritten again, please. Please make it more ... highlight one point more than the other, or maybe make it shorter.” Sensational feature! You’ve got a text that’s 30 pages long, and you say: “Please, please, I don’t have time - no joke - summarize it for me. In two paragraphs, and pull out the top three points.” Snap! Five seconds later you’ve suddenly used “generative AI” to save yourself a lot of time.
Isabell: I think that interaction is what makes it special, that you can really interact with an output once more.
Christian: That’s sensational! I mean, I think the people who, for me personally but also in public, were the first to really bring it out. That was when, for me, at the beginning of last year for the species that were still on it earlier, already in October before that, “ChatGPT” suddenly came out from “OpenAI.” And they really set it up exactly like that too. So this language capability can be used in a thousand places, but the moment you build it into a dialogue, it becomes a completely new experience!
Lisa: And now let’s translate that into the things it actually gets used for at eBay. Your title says you work on the marketing platform, but beyond that, what initiatives or topics at eBay involve AI?
Christian: In general, I’d like to divide that first into three different areas that you can look at qualitatively. One is how AI can be used to optimize internal workflows and processes within eBay so that they simply become more powerful and more customer-friendly. These are things that people probably don’t necessarily notice from the outside right away. Things like using AI very nicely to write much more program code much faster. As a customer, I may only notice that indirectly. But if developers can suddenly produce or deliver new features much faster, then that is of course an advantage for me. Or customer service, where AI is used to support the employees on site, perhaps. So these are internal things that you as a customer don’t necessarily have to notice at all. The other is very, very concrete. The second category for me would be something like: how can we use AI to really improve the user experience of the user? And of course that works wonderfully on both sides, both on the buyer side and the seller side. In other words, what can AI do to make the shopping experience on the platform even more pleasant, even better? Or the selling experience? We may come back to that in a moment when we talk a bit about concrete examples. So that would be the second pillar. And from my point of view, the third part is again more invisible or only felt very indirectly, and that is the question of how AI can be used to basically achieve the actual purpose of the marketplace: bringing people who want to buy and people who want to sell together. Where can AI be used to make that even better? So to connect them even better, to help one find the other, for example.
Lisa: I’ve never understood it this well before as I do right now. Thanks, Christian.
Christian: Too kind!
Isabell: Okay, so AI improves the selling experience. What does that mean specifically for our sellers?
Christian: First of all, for everyone who may now be thinking, oh my God, there’s this huge AI wave and it’s going to knock me over, and I’m missing all of it and so on. AI is fast and, for sure, will change many areas. But I’d rather call it an evolution and not the big revolution where you wake up tomorrow morning and realize, oh look, eBay has suddenly become something completely different. And that’s how I’d understand our approach here too. We’re continuously looking at what developments there are, what technologies exist, what potential each one has. And then we try to build concrete, and at first also quite small, building blocks from that, AI building blocks, so to speak, which we then let flow into the product.
Isabell: And what building blocks are we building right now specifically? Is there anything already that sellers are benefiting from today in this evolution?
Christian: So, um, yes, of course! When you think about where AI can be integrated and where it can be used best, you usually look at areas that seem complicated, time-consuming, and difficult, and you check whether there’s anything where I can use AI to make this problem easier to solve with the possibilities we just discussed. And I’d also interpret our roadmap in this area in that direction. For example, a feature that is already available in Germany for all sellers is the ability to generate an item description with AI. Anyone can try it right away. You’ll find it in the app, for example, with this icon that has these typical little AI stars on it. If you click on it, it generates something. And the response we get to that is extremely positive, because it’s very simple, it saves time, it helps people, and in principle you can then create more listings much faster and ultimately do more business on eBay.
Isabell: Yes, we also often hear from sellers that they don’t use it to replace their own texts, so to speak, but to supplement them, because they’re nice texts too, so they use it more as a source of inspiration.
Lisa: It’s almost a bit like a synonym search sometimes. I used it privately and was totally amazed by how eloquent our AI is.
Christian: Yes, unfortunately there’s nothing to add to that, I’d see it the same way. And I think it’s a very nice example of how you start using AI without suddenly having completely changed everything. It just makes things more pleasant, better, and therefore very helpful.
Isabell: Are there any more tools we already offer?
Christian: It all goes in that direction. Another classic is: oh no, images. My image isn’t good enough, or I need a different image. That means a lot is happening there too. For example, how can AI be used to replace or remove the background from an image?
Lisa: Also super nice, especially when we think about the fact that a background, for example, must not even be there for third-party platforms in order for us to be able to advertise those items. If we give our sellers a tool that simply cleans up the backgrounds, I’d say that can also just be an increase in sales.
Christian: Absolutely, and above all it simply removes a barrier. Because if I’m a seller and I’ve got, say, 1,000 or however many items, listings on the platform. And then I see, oh yes, I need to remove the background. I know, okay, maybe I’ll do that the day after tomorrow or something. But with our friend, friend AI, you can possibly do something like that much faster, and that’s in everyone’s interest.
Isabell: Yes, and above all it’s then just one click and no longer this erasing by hand that we remember from the past.
Christian: That’s a nice example and you can also, of course, take it further. Something that’s being worked on, but isn’t yet available in Germany, is for example the idea of creating an eBay listing directly from a photo. And for me that is dreamland. I have something lying around or standing there that I think, yes, I could actually sell that now, but do I have the time right now? Then I just take a photo, and boom I have my listing in the system and can sell immediately without spending any more time on it.
Isabell: So the AI then automatically fills in the category and the item specifics and probably uses the AI-generated item description directly as well?
Christian: Exactly, that’s used too, exactly! And this feature itself is, I’d say, currently in some kind of beta phase. So it’s still being tested. And what it will look like in the end, very, very exactly, and where and when it will be available, also depends a bit on what other insights come out of it. But that’s exactly the direction it’s heading in. The idea is: I see something and want to sell it, and I don’t want to deal with the details of creating the listing, but rather have the AI do that for me.
Lisa: Super beneficial, then, for every seller listening right now and thinking about how many product managers or product managers might have to be consulted to find the right description or the right title, because you need to know the product well enough, that you may not need that anymore and can use resources elsewhere, because listing on the platform is simply so clever that you can then use the good internal resources more for product sourcing. Or something like that. Huge time saver!
Christian: Yes, and you can already see that experienced sellers will now say, yes, okay, but what if the AI sees something wrong and suddenly the wrong version of an iPhone is listed and goes out at the wrong price? And that is exactly why I also expressly say this is something we are currently looking at, experimenting with, and which is not yet available broadly. Because the devil is really in the details, and of course we absolutely do not want to make the process easier but then introduce errors. So there is still a bit of work to do there, for sure, but very promising and I think that direction is definitely worth pursuing.
Isabell: Super exciting! Now, we’re also using it in the ads area, can you tell us a bit more about what exactly is happening there right now?
Christian: Yes, that’s actually also a classic use case for AI. A very nice field for AI, because if you want to optimize ads, and if you don’t have a lot, a lot, a lot of “know-how” at that point, it can very quickly become very time-consuming. Because you have to look at the different data sources. You have to think carefully about the strategy and how you want to structure it on eBay. And for something like that, AI is really very well suited. Looking at large amounts of data and understanding what this ad is about, for example, and who might see it, or in which channel I want to place it. That’s a very good topic. So that’s already been running for a long time. The unfortunate thing for us here today is that this functionality is basically in the background; it falls into that third category I mentioned earlier. What can we do to improve it? But you don’t really see that much of it, unfortunately!
Lisa: But maybe sellers do feel it in the results.
Christian: I’m very sure they do.
Isabell: I was just about to say, the news isn’t actually that sad if I know that what I’m seeing is already basically AI-based.
Christian: That’s true, of course for me: then nobody even notices it, so you can’t really celebrate it. But yes, that’s true, the perfect AI would be something we wouldn’t even notice, it would just be in the background and simply make our lives nicer and better. But we’re probably still a few steps away from that.
Isabell: Where we can still see or notice AI or use it consciously is in the social media area. That’s going back toward “generative AI” too. What exactly is happening there right now, Christian?
Christian: Also a very nice field for AI, because everything is also a question of time. You had a listing, and then you want to promote it or post it on different channels. And then you’d actually always have to think carefully: wait, which channel is this now? And how do I do it now? And do I really have the time for that? So it’s probably more of a nuisance in many places, and maybe it often just doesn’t happen at all. And AI can help wonderfully there. Telling an AI: “Look here, this is my listing and I want to put it on this channel,” then the AI generates a very nice post for it. And then the seller just has to use it to post it.
Lisa: Christian, you talked about this wave that can kind of overwhelm or overtake you a bit. And AI is already a topic that makes many people uneasy or worried simply because they’re not so close to it. We stand for the responsible use of AI, but what does responsible use of AI actually mean?
Christian: Well, there were really two points in that. And I’d also like to say something briefly about the first one. Namely, for everyone who opens their email in the morning, Twitter, or whatever channel they use to get informed, and tries to avoid the topic of AI because every day it feels like they’re being flooded and oh God, I can’t read all of this and I don’t even want to. But actually I should, I can only say: first of all, I don’t know if it helps, but I feel exactly the same way. Even though I love dealing with it, it’s a lot of fun, I still often enough, I’d say several times a day, feel like, really? Already this fast? So don’t worry, that’s completely normal and completely human, and that shouldn’t stop us from dealing with it. What’s important, and I see it this way too, is that we are completely clear about how we use this technology. Because the possibilities are, I don’t want to say unlimited, because that would of course be nonsense. But there are a lot of scenarios in which AI can be used. But for us as eBay, it has always been extremely important from the very beginning that we use certain ethical principles when dealing with AI. And there are also, online, various sources. I think that’s very high quality, of course, but also as a private person it speaks to me strongly, and I’d like to mention my top three, just briefly here, and also pass them on to everyone who wants to use AI, no matter in what context. It’s really important to realize that there are risks and that you actually have to think beforehand about how to deal with them. It is not a good solution to take your customer database and throw it somewhere into AI to see, yes, what else could I improve there or something? Because once the data is in there, it’s in there. And to slowly return to your question, dear Lisa. My top three AI ethics points are: One, or the most important one for me, is to ensure that an AI is not biased. That sounds totally crazy now. Why should an AI be biased? It’s actually just a computer program or something, right? But you have to understand that an AI is nothing more than a certain algorithm that has processed, let’s say, unimaginably large amounts of data in order to learn from it. So, for example, if there are clouds in the sky, then it rains, no idea, just to take a very simple example. But this can go arbitrarily deep. And then there are truly awful examples. For example, you tell an AI: “Please generate an image - and I’ll make a few examples now - of a manager, doctor, prison inmate, model.” And if you look at that, what will come out, for example, you absolutely have to do this with Nobel Prize winners for a thing like that. And without commenting on that further, when ten images are generated, then with very high probability nine out of the ten images are old, bearded white men who then get a Nobel Prize.
Isabell: And how can I avoid that?
Christian: Only the person who essentially controls this process can moderate that; it’s called “training.” So this AI is, at the beginning, basically stupid, and then it gets training material. For example, images — it gets images of scientists, images of prisoners, images of this and that, and learns. And then it understands, very simplified, aha, that’s a typical prison inmate or a typical tree, a car, whatever. But if you just give it random images without control, then the AI will afterward simply reflect back what it found in the images. And there were, this was a few years ago now, but there were attempts by quite well-known “players” too. They went ahead and said: “Oh man, great, we have an AI, let’s switch it on and let it learn online.” They simply read news sources and followed tweets. And I think it took no more than three or four days before they shut the thing down again, because it was highly racist, misogynistic. It simply absorbed the mood of what it learned from. It’s basically like a child: if you take a child that is born completely neutral and it grows up in a society that is not open, transparent, fair, or diverse, then it’s very likely that the child will also develop corresponding values.
Isabell: Okay, so to tie it back to the principle you just mentioned: it’s important, so that the AI is unbiased, that the people feeding the AI information also essentially “monitor” that information and don’t let it run out of control.
Christian: Yes, I mean, that topic alone could fill an entire evening or even weeks, if you will. But in principle, yes, that’s right. And where that doesn’t work, I can’t, not everyone can train their own model, because that’s a huge amount of work too. But then those who use it have to make sure that it is not, so that this is taken into account when using the model. Super important, and there’s no excuse there either. Everyone who uses AI must first understand this problem and think about mechanisms for dealing with it. The best thing is to do the “training” properly. So that would be the first point. Super important: when we use AI, no matter where, whether I’m doing a product search at some point or in customer service or wherever. A person deserves to know whether they are interacting with a human or with a robot, computer, or AI right now. That means this must be a basic principle: when we use AI, we make very, very transparent where we use it, how, and what we access with it. Important basic principle. And the third point is that when dealing with AI, you always have to look at how we protect the user’s privacy. So you very quickly end up in a chat setup, not necessarily, but it really lends itself, as we discussed. And in such a chat, as the provider, we have only limited control over what is entered into that chat. And it’s important to make sure that nobody accidentally enters data there that shouldn’t be entered there at all. So that would also be a third important point.
Lisa: Super important to hear the basic principles for your own behavior, and it’s also very, very nice to see that eBay puts a lot of thought into this. Lisa: And if you want to hear more about these principles, of which Christian just listed three: they are public, and we’ll link them for you in the show notes as always. Coming up, we’ll return to our question from the intro: Is artificial intelligence revolutionizing shopping behavior and, with it, commerce? But first, we want to hear from you out there. Are you already using the AI tools to create item descriptions, item images, or social media posts?
Isabell: And if so, let us know what you think of the tools. If you’re listening to us in the Spotify app right now, you can also do that directly via the “Q&A function” in the app. Just drag the “slider” from the bottom edge of the app upward and off you go.
Lisa: If not, feel free to write to us in the eBay Community or on our Facebook page “eBay for Business Germany.” We also welcome your feedback on the podcast and, of course, a like or a subscription where you’re listening to us right now.
Isabell: So, Christian, now back to the question we’re all so eager to know: Is artificial intelligence now revolutionizing shopping behavior and, with it, commerce too?
Christian: You always have to be a little careful with the word “revolution.” Will commerce change because of AI? Absolutely! Will it happen overnight? I don’t think so. I think if I were to speculate a bit about where this could go, then I’d simply say that in one, two, or three years we’ll all have gotten used to surrounding ourselves with these kinds of “AI assistants.” That will be completely normal for us. It will simply be in contact with us in our messenger apps. One of them will be a fitness advisor for me. It can then say: “Hey, you need to go running again soon!” So for all kinds of life topics, that will be completely normal. And those will be AIs, and they’ll advise us like that, and logically there will also be AI assistants taking care of the topic of “shopping.” And I actually believe that user behavior will definitely change over time because of that. We’re used to searching via search engines - whichever one it is. And that means we enter something we know we need, and then we get search results. And then we decide whether we want to buy them or not. In an AI-driven environment, it will be more of a conversation. That means I’m now talking to a virtual assistant, an AI, and discussing with it what I actually need. And no longer: “I need a concert ticket for concert XY.” But rather something like: “My brother-in-law, whom I only see twice a year, who likes sailing and is turning 50. And I have no idea what to give him. Please!” And then the AI says, probably the AI will ask questions and say: “Yes, sailing is a good start, what kind is it? What are your favorite colors or whatever?” Then you’re suddenly in this dialogue, and in the end you’ll still, strictly speaking, get something like a search engine result, but much more tailored and with much smarter hits. And that will be, that will already be a new, that will already be a new shopping experience, I think. The other thing I’d expect is that classic product purchasing will increasingly merge with other activities. So, for example, today if you’re planning a birthday, you’re planning a birthday party. Then you have various activities, like picking music, inviting people, preparing a buffet, and however else you celebrate parties. And that always includes buying products too, for example. And today that’s traditionally separated. That means you do your party planning and then you pick the people and then you write an invitation and whatever else you do. Right, and at some point you also order some products. Fine! In the future it could be that you’re actually talking to your party bot and saying: “Yes, I need here a, I don’t know, birthday for a child of such and such an age or for me or for whoever.” And this bot then takes care holistically of what you need. So it writes the invitation card for you, creates the playlist for it, figures out which people you could invite, and so on. And I know, Lisa, you look horrified. Oh no, how can the AI possibly know all that? I know, that brings us to data privacy. But that means we’ll still have to see exactly how that will work in detail. But in principle, it will be like this: we could all have a kind of guided shopping experience where an AI takes us by the hand and helps us get exactly what we actually need.
Lisa: What you read on my face was more fascination, because I’d love to have more chatbots for all the topics in my life.
Isabell: And I find it funny that you’re using the birthday party example right now, because I actually just used AI to plan my birthday party. Especially with quantities, what could I make to eat, what’s there to drink? I invited X people and so on. Definitely super exciting if that ever gets combined with: “Buy it directly here!”
Lisa: We recently had it calculate how much “crushed ice” we’d need to buy for a party with 100 people, of whom how many are buying alcoholic and non-alcoholic drinks. So you’re not far off!
Christian: Wow! That’s very good, very nice!
Lisa: Christian, if I’ve listened carefully to this episode and I’m a seller on eBay myself, then I’m a little left with the question, “what’s next?” So leaving this episode, what do I do now? Do I wait until eBay comes to me with information? Do I keep myself informed? What should our listeners do now?
Christian: A bit of everything you just said. What I would absolutely recommend to everyone is to definitely keep track of what’s happening right now. Strictly speaking, no matter on what level. You don’t have to spend hours every day dealing with AI and what’s coming. But if something comes through the media, take a look and try it out yourself. So your example of planning parties or ice consumption or whatever, excellent! Really good, because you suddenly get a very personal feel for it and can then much better judge whether this is something for me in a certain situation or not. The big problem for sellers is of course that you already have plenty to do, and becoming an AI expert on the side is difficult, and I wouldn’t even recommend it, because it makes no sense if 100,000 people do that. And that’s the point, and maybe that’s also my main message here today for everyone listening: eBay, as a partner to commerce, is also a partner to commerce here. That means we really have a lot, a lot, a lot of attention, focus, and activity centered solely on looking at what’s happening right now, where it’s going, what’s useful, and what may not be useful at all. There are also things that pop up, get totally hyped, and then disappear again. Too bad for everyone who spent time on them, because it was basically wasted. We can wonderfully cushion that. So being on eBay as a seller is definitely a good ticket on this AI train, because we make sure that what is sensible and necessary and possible is made available in a way that is ethically clean, reasonably developed, and well managed.
Lisa: That was a very, very nice summary! And attentive listeners will know what comes now. Because, Christian, at the end we always ask: what was the last thing you bought on eBay?
Christian: Yes, I, hmm.
Lisa: Did an AI tell you what to buy?
Christian: That would be nice, that would be good! Then it wouldn’t be, I mean, lately I’ve bought, I checked again, there was a lot of toys and books involved. And that’s all ultra boring. But I need a new bike. That means the prediction of what I’m going to buy next is much more exciting than what I’ve bought, and that will be a bike.
Isabell: An e-bike or a regular bike?
Christian: No, I mean, I love e-bikes, but for health reasons it wouldn’t be an e-bike.
Lisa: Then let’s see what you’ll be coming to work on in the future. And with that, we’re already at the end of our episode. Thank you so, so much, Christian, for the exciting insights into the world of AI and for the many, very, very lighthearted explanations. It makes us super, super excited for the future. We’re looking forward to everything eBay will still provide there. Thank you so much, Christian!
Christian: Thank you very much, it was fun.
Isabell: Thank you from me too.
Lisa: Goodbye!
Christian: Bye, tschüss!
By
5 comments
olavgo13eba
·8 months agoA simple search today shows that several living animals, such as puppies, have been offered since at least the beginning of the month without any human staff member or AI having intervened. The eBay robot even overlooked several phone numbers mentioned and did not remove them.
vierterplanet0
·8 months agounfortunately, the employees on the hotline now already sound like they themselves are artificial intelligences. 😞
olavgo13eba
·10 months agoThanks for your opinion @michiskleinkram. I really can't follow your argument, because, for example, even impermissible search ads (see the relevant eBay policy), which can't generate any sales and usually only frustrate potential buyers, remain online for years …
michiskleinkram
·10 months agoYes, unfortunately that's how it is. The staff apparently are under pressure from Ebay to let the listings "go through." Ad revenue is everything, after all, unfortunately!
olavgo13eba
·1 year agoIt would be nice if eBay would finally use the announced AI tools first and foremost, so that prohibited items like living animals (dogs, cats, birds, rodents, …) don’t even go online in the first place. Bad enough that, for example, dangerous snakes offered there are always declared permissible by eBay employees even after multiple reports. Clearly eBay has no interest in actually putting into practice, for example, its supposed commitment to animal welfare (see eBay principles), does it? Not to mention the server capacity consumed, the manipulated search results, energy consumption, etc.