LinkedIn's War on AI Slop | Claude's going for Privacy | Tech Jobs Report

Hello everybody and welcome back to our weekly video where we cover everything that's been happening in tech, data, and AI. And in this week, we don't have a ton to cover, but there are a few really interesting things to look at today. The first thing is that LinkedIn is finally taking some steps to counteract some of this AI slop that is everywhere on the platform. This right here is Harry and I'm going to butcher this last name, but it's Srinivasan. He is the chief product officer at LinkedIn and so he is talking a little bit about what they're trying to do. He says right here, AI slop is a top priority for all of us. We really care about this and people come to LinkedIn to connect with real people and share their real perspectives, ideas, and expertise, which is uh in my opinion 100% true. I don't really like coming on LinkedIn and I post and I get lots of just AI responses. I do not like them. They are not really engaging to me, right? These aren't real people who are actually giving their feedback. It's just you can tell immediately that it's AI slop. Now, I'm just going to give you one example. This is a post I made uh several days ago just about, you know, a silly story about a printer that wasn't working in an old office. True story. I wrote it and within minutes, there were already comments like this and people like it. I don't really understand why. It says, "The no one complained, they just figured it out part is the real story." I mean, anytime it starts out like this, it's it just sounds like AI. Maybe this person really took their time to write this out and write it just like this, but I'm telling you, it it I can spot it in a fraction of a second whether it's AI or not. And there are ones like this where they're just like random channels or stuff on LinkedIn and these aren't real. And so this is what they're talking about. Now, it's not just comments, although that's where I personally see it the most. It's in just people auto commenting. They're basically summarizing saying, "Hey, AI, write a nice response to this on LinkedIn." And they give the same generic output every time, but it's also posts. It's posts as well of people just posting random AI slop that is just garbage. And so they are working on several different things. And one of the things right here is they're in beta version for this right now, but you're going to be able to take a look at a post, take a look at a comment, and you're going to be able to report it and say, "Hey, this seems like AI slop." Ironically, most likely what is happening on the back end is you're just training their AI in order to detect the AI better. That's probably all we're doing. It is possible some better system because LinkedIn is, you know, pretty good network overall. So, they're going to have things like machine learning systems in place to detect these things. So, it's possible it's not AI. I'm just saying it would be really ironic if we are just training an AI system on the back end for it to then detect AI while we keep identifying this as AI. I just think it'd be ironic. That's all I'm saying. Now, there is some good news. It says, "On comments alone, every day we're now catching hundreds of thousands of automated comment attempts and have blocked billions of other automation attempts, posing slop in the last couple months alone." Now, I am not shocked by this because there is so much opportunity here to very easily get rid of just so much junk. I'm glad they're at least making progress towards this. It also says that they're ramping up a series of new and improved classifiers that identify if a post is AI slop or generally low-quality content. So, even if you aren't using AI, if your content is just not good, right? You're just writing random junk and it doesn't sound good, it's not, you know, good grammar or just sounds like AI. It is possible that they're just going to classify it as low content and not show it to a lot of people, which I'm okay with. I'm fine with it. If you're not putting a lot of effort into your posts, if they're not actually good posts that people will want to read, um I'm okay with them getting identified in this way and then just not being shown. That's okay. And then right here it says, "We're ramping up the ability for members to tell us if they believe a post or comment seems like AI slop." Slop is hard to define and the definition changes, but they're tuning their models to make their better feeds. This is what we're talking about earlier. I think this is great, but again, this is purely done by people, right? They're identifying this by you and me saying, "Hey, this is AI slop." Um there's a lot of ways this could go right, a lot of ways this could go wrong. We will see how this plays out, but giving people at least the ability to be like, "Hey, I don't like this. I don't want to see this type of, you know, comment because I'm pretty sure it's AI." I think that's a good thing overall. And one thing I do like, and this is because I post a lot on LinkedIn, is for anyone who shares content will test a way to privately flag in your analytics dashboard when members feel your post may have come off as inauthentic or heavy use of AI. So, this is good for somebody like me because if I'm making posts and people are flagging it in droves that it's AI, well, then maybe I'm just bad at writing, and I need to get better at it, and that's fine with me. Maybe I'll need to work on this. But, it at least gives people the opportunity to say, "Hey, people think my stuff is AI, the way I'm writing or the way I'm creating it seems like it's AI. Maybe I shouldn't write like that anymore." It doesn't mean you have to change. It just means that that's part of your algorithm now. You need to be aware of this. The next big news is that Claude code can now run its own compute. Now, if we come right down here, it says this is in public beta, but you can self-host environments where you can run Claude code sessions on your own infrastructure. Now, just at a high level, this is amazing because up until now, you cannot do that. And so, everything that you were doing had to go through actual Claude code, and with this change, here are some of the reasons why you'd want to self-host. One is that sessions run inside your network and can reach internal services, databases, and registries without exposing them to the public internet. So, everything can be run on your local computer never having access to the internet, which means you can't really accidentally mess things up. Now, once you connect back to the internet, if you have, you know, bugs and issues in your code or, you know, API keys or whatever it is, then yes, it's still going to, you know, mess things up. But, if you're doing everything locally and you don't need to connect to the internet, you are good. And that is a huge, huge benefit of doing something like this, especially if you're already in the cloud code ecosystem or that's kind of part of your workflow. You can also pre-install compilers, SDKs, and internal CLIs in your environment, so every session starts ready to build. Now, this is awesome just because again, you don't have to go through and make all these connections and kind of log in to all these different things. You can have all of those things already pre-done on your computer, so you don't have to then, again, configure everything when you want to go and work. And lastly, your source code and build artifacts stay on the infrastructure that you control. I think this is just the way that AI is going to be moving a lot in the future, especially for professionals who are working on, you know, things that maybe you don't want to be sharing with the outside world, really sensitive data, some proprietary code. People want control over what they are sharing and what they're not sharing, and some of these things are really private. They don't want to accidentally, you know, share all these things, but it happens a lot. And so, that piece of things, especially for, you know, newbies and just people getting into AI, they don't know all these different things, that is a great use case, you know, just having that kind of boilerplate AI. But, for professionals who are using this for their work and it could cost them a lot of money uh to mess this up, this is a fantastic thing. You're going to see a lot of startups going this way. I just think for, you know, your average AI user, they're not going to care about this at all. But, for developers and software engineers, even data engineers, people who are working really heavily with these AI systems, and this is a big, big change. Lastly, I want to touch on the job market. Um this is from CompTIA. Uh I'll leave a link in the description, but it basically just summarizes some of the information from the Bureau of Labor Statistics. And if you've ever been on the Bureau of Labor Statistics website, it's not the easiest to read. I mean, I went through and read it, but then if I want to show it to you, it doesn't read like it does in this article. And so I just pulled up this article. Um but basically, the technology sector expanded by approximately 3,700 jobs in July. That's not a lot. Uh so that's not great. But there is some other data in here that I think is pretty interesting. It says employer demand for new tech talent remains strong with nearly 603,000 active job postings nationwide and across all industry sectors. If we look right down here, it says demand for artificial intelligence AI skills remained significant with nearly 14,000 job postings in July alone seeking AI and machine learning expertise. At the same time, employers sought out talent for established tech roles at a much greater scale, including software engineers with 46,000 job postings, systems engineers with 34,000, tech support specialists at 24,000, and data analysts at 18,000 and 800. So there still is a lot of demand for these jobs. There's a lot of job postings. What I have anecdotally been seeing, and I absolutely want to, you know, make a full video diving into the data on this. Anecdotally, I have been seeing that there are a ton of job openings, but there are so many people applying to a very small number of these job openings that it's really difficult, right? So especially the remote jobs, you're going to see 10,000 people apply for a handful of remote jobs that are available. Whereas the local ones are going to get like 50 applications. And that's purely just because they're local, they're most likely not going to be making as much money. And so people don't apply to those as much. Anecdotally, like I said, I have talked to so many people who have gotten jobs in the past year where it hasn't been the best job market. And the majority of them, I would say 75% of them have said that the jobs that they got were local jobs or ones that they had to move to be in person for. They were not remote jobs. So, if you're just starting out, I highly recommend taking a look at your local jobs or somewhere where you can move to be local. If I'm living in the middle of the mountains in North Carolina or South Carolina and there's Charlotte, you know, and there's Charlotte, North Carolina, which is a big, you know, metropolitan city and there's lots of jobs there, then I can apply there and if I get a job, then I can just relocate very easily just several hours. And so, if you are living even within a vicinity where you are open to moving, I highly recommend it. Now, that is all we have for this week. I think the LinkedIn stuff is great. I genuinely think it's a big problem on the platform. I think they're aware of it and that is a really good thing is just being aware of it. I myself get so frustrated when I have 20 comments on a post and, you know, 15 of them are AI. It it just isn't fun as a creator to try to interact or parse through which ones are AI slop and which ones are actually real people who I want to connect with. Cuz that's kind of a big reason why I post. Cuz I like talking to people. I like connecting with people. But when most of them are just people using some AI system to auto comment, it's just not fun. Uh that doesn't excite me. Uh so, I am glad that they're taking this, you know, really seriously. That's kind of the biggest thing or the biggest takeaway from, you know, this week is that they're at least trying. All right, not every platform is going to be great at it or get 100% but they're at least trying and identifying it. Uh the code thing with uh Claude is amazing. I think this is long overdue. I'm glad it's happening and I'm glad that we can all now benefit from keeping things internal instead of having to keep everything connected to the internet and expose, you know, privacy things that we don't want to have, you know, accidentally slip out there. And so, those are all really good things. I hope you learned something from this video. I'm releasing these every single week. So, if you like this type of content, be sure to like and subscribe, and I will see you next week.