Speaker 1: I can't believe it. A politician has been caught lying. I have never heard anything like this before. This time however it is a scientist slash politician that has been caught. They've been caught fabricating data in a really interesting way that I'm going to share with you using AI tools and the villain of this case is the Italian health minister and so they have been caught using software that I've never seen before but I'm very excited about. I've even given it a go to stay around and you'll get a chance to see what it actually can do. It's very very powerful but this health minister has got a background as a scientist so if you look here you can see that they have got a massive h-index of 49. They've got a load of citations and these are all of the papers they've published recently. They publish a lot. Look how many. There are loads and loads of papers with their names on it and also for a lot of them they are corresponding author. Now that's very important because if you're corresponding author it means that you can be contacted about the information in that paper and it's the most prestigious position really in an author order. There are so many papers in that list but you can see that they all follow this kind of same structure. You normally have some sort of introduction with an animal model but importantly the thing that they have been caught out about are these sort of images with cell microscopy images. Now it's really hard to try to find similarities between microscope images because the resolution can change, the image can be manipulated but I decided that I would just check a couple of these papers and I'm amazed that me with my human eyes can detect where they've just reused images and it is super easy. Let's take a look at these two images. So this is from one paper figure four and this is figure one from another paper and I was just going through and looking for images that were the same and here we have it on figure four B and ML. They're exactly the same image and those scale bars are the same size but you can see that this is a much more zoomed out image and they claim on one of them that it's prostate cancer cells and on the other one it's breast cancer cells. So clearly that cannot be the case. You can see it a little bit easier with my super awesome image analysis skills paint. I used paint but you can see that this little E here is this one here. You've got this little dark patch. Easy, super easy to notice but then I was looking at other papers that apparently had duplicated images and I was like okay let's see if I can use my awesome eyes to do the same thing and ultimately I couldn't. I was looking through this paper and I was like I cannot work out where the duplication occurs. Apparently there's the same image repeated in one paper so that's when I decided to use ImageTwin. ImageTwin and this isn't a sponsored post. ImageTwin is something that I've never seen before and I'm incredibly impressed with its powers. It uses AI-based software for detecting integrity issues in figures of scientific articles. Check for plagiarism using a database with over 21 million images and they look for image manipulation. It's really simple. I was surprised. You upload your PDF, you scan it and you get a chance to see what the results were. This is the result from the paper that I could not detect with my own human eyes the duplication and it all comes down to this figure here. You can see in these boxes that is where the image duplication took place. Let's take a little bit of a look here. This is the full image but let's have a look at the area where this image duplication takes place and you can see that it's a pretty solid match. That is something that I wouldn't be able to do with my own eyes and particularly if you're looking at these papers day in day out there is no way you'd be able to see that this and this was from the same sample. It is almost impossible to detect what by just looking. So this image software has done an incredible job and I think this is just the tip of the iceberg of where this kind of like fake data and data manipulation technology is going. Very impressive. The politician is actually quoted as saying this is the first time I'm hearing of this news. I had no knowledge of it. I am not an expert in electron microscopy. I trusted the person who provided those images. We will check to see if there are indeed errors. Now that is not good enough. If you are the corresponding author on a paper you need to make sure that you are comfortable with all of the images, all of the data that is being represented. As the corresponding author, if you are publishing so many papers every year that you are incapable of checking the quality of those papers, you are publishing too much, you're not doing your job properly and just saying oh I don't know I trusted someone else is not good enough. These sort of supervisors want all of the glory but do not want to accept the responsibility that comes along with being corresponding author. And we can see here that the rules of the scientific community dictate that those who lead a team of researchers and put their name to the publication bear the onus of ensuring their authenticity. Exactly. Now this isn't the first time this has happened. We got German politicians that are dogged by claims. This was in 2021. There were all these politicians that were seen to have fabricated their thesis, had plagiarism issues and I was looking deeper into it and it was pretty interesting because apparently in the past to be able to get into politics in Germany you had to be part of the aristocracy. But there was a way to shortcut that and that was having a PhD. If you had a PhD you were able to become a politician and that's why we're seeing all of these people in 2011-2013 resign and lose their PhDs for making it up because they essentially just plagiarized to be able to become a politician. In an article inside Inside Higher Ed in Pakistan the constitution required the candidates for public office to hold university degrees and so fake degrees were just rampant. They were just paying for degrees. The one thing I love about this is that one of the Pakistani chief ministers said a degree is degree it does not matter if it is real or fake. Why do we even do this then? Just start making fake degrees they're way easier. So why? Why does it exist? Well ultimately I found this recent paper in the Journal of Mental Health and it's about the anxiety of the lone editor. The lone editor for journals where the companies that quite often sit above them and employ them as editors they earn billions of dollars. They do not give them enough resources to be able to detect fraud, detect fake data, detect paper mills and plagiarisms all that sort of stuff. This is a pretty interesting read so I'll leave a link in the description. Essentially they're just saying it requires 24-hour attention to prevent untruths and lies getting into the publication record and that takes a toll on editors. After all they're only human which is why we need to start using AI for this sort of stuff. Their vigilance requires a knowledge of the cues of wrongdoing and keeping ahead of those trends is time consuming leaving editors exhausted and stressed in case they let something through that taints the scientific record. They often use tools like similarity check and authenticate but ultimately I think this is a sign of how AI tools can be used to detect fraud before they even hit the scientific record. And I think that in the past tools were only available to publishers. I think now as a scientist it is up to you given the amount of fabricated data there is out there to be able to use a tool like image twin or whatever comes out in the future. Using these tools will give you the power to actually decide whether or not there is fake data. If something seems too good to be true it may be worth a little bit of money from your research group to pay for a tool like image twin just to sort of like do that last little check because the last thing you want to do is base your research on this thing that is clearly false and fabricated. You can waste a lot of time and so just spending that small amount of money can really just help you avoid those pitfalls. It's quite expensive it cost me 25 euros for one scan but if your institution is serious about making sure that you don't end up just like wasting your time chasing something that will never work because it's faked in the literature this may be something you'll be able to convince your supervisor or your school to pay for. If you like this video remember to check out this one where I talk about fake scientists infiltrating top journals. That one is a great watch. So there we are there's everything you need to know about the recent developments in AI tools that allow you to detect really complicated fraud and fake data and the Italian minister's fake data scandal. Let me know in the comments what you would add and also as always there are more ways that you can engage with me. The first way is to sign up to my newsletter. Head over to andrewstapleton.com.au forward slash newsletter. The link is in the description and when you sign up you'll get five emails over about two weeks. Everything from the tools I've used, the podcast I've been on, how to write the perfect abstract and more. It's exclusive content available for free so go sign up now and also go check out academiainsider.com. That's my project where I've got my ebooks, I've got my resource pack, I've got the blog, I've got the forum and everything is over there to make sure that academia works for you. Alright then I'll see you in the next video.