{
  "video_id": "y3DQZHw7b3w",
  "channel_slug": "themorpheustutorials",
  "channel_handle": "themorpheustutorials",
  "title": "Boost Your Cybersecurity Career with AI Mastery!",
  "duration_seconds": 2164.0,
  "url": "https://www.youtube.com/watch?v=y3DQZHw7b3w",
  "upload_date": "",
  "transcript": "AI systems are getting more and more powerful.\nClaude, Gemini, ChatGPT, they program things for us, they write reports, they write policies.\nBut when you listen to the IT, you always hear one thing.\nMainly the older semesters, those who already knew quite well without AI, benefit from it.\nThey can use the whole thing as an insane speed-up and work much faster and more efficiently.\nJuniors, who are relatively new to IT, or people who want to use AI for learning and want to get into the field,\nthey have difficulties with it.\nThey have it more difficult on the job market, because you need the seniors, because they also want to deal with AI.\nAnd at the same time it is often very difficult to find the topic at all.\nSo let's look at it today.\nHow can juniors really use AI in a meaningful way?\nAnd how can juniors really use AI in a meaningful way?\nWe probably make several videos out of it, depending on how you find the whole topic.\nAnd start with cyber security.\nThat's actually my area.\nPlus, of course, AI. I do both.\nOr I've actually done both before.\nAnd we want to see how exactly you can use AI as a beginner, as a junior and then later as a senior,\nreally use it sensibly and efficiently and not accidentally fall into a trap.\nAnd to do this, let's take a look at exactly this.\nBecause my observations are, when I look at your messages,\nwhen I look at the messages from older colleagues,\nand when I just look at the students that I sometimes also look after,\nthen this is the following.\nIf you use AI without really having an exact understanding of what exactly will be done or what you want to do,\nthen you will of course make worse inquiries.\nMy inquiries, for example, will be much worse in the area of,\nbecause I can't speak the technical language at all, than that of a lawyer.\nLogical, isn't it?\nAnd of course it is the same in IT.\nAlso the output, of course, I can evaluate or use much worse.\nBecause I often don't understand them at all.\nAnd that leads to a problem.\nBecause if I use AI, I have to rely on the output of AI and thus learn less,\nbecause I don't even get into the topic.\nAt the same time, it is expected of me that I can do more,\nthat I can work faster, because I have AI at my disposal.\nAnd so it has to go faster.\nThat is exactly this problem that we hear over and over again.\nAI doesn't make everything faster.\nIt makes the activity that you really should do at a certain point of time faster.\nAnd that is fair enough as a beginner and as a junior\ndefinitely not yet with AI to do everything and then just read over it briefly\nto see if that's really okay.\nThe next thing I notice is that people keep writing to me,\nhey, AI makes mistakes and I can actually do it myself.\nAnd that is also in a way right.\nBecause if you don't deal with the outputs properly,\nif you haven't dealt with the topic properly and haven't worked in depth,\nthen I can't even see if the AI makes a mistake.\nBy the way, that's exactly the problem with vibe coding.\nYou don't know your own code anymore.\nAnd if the AI really makes a mistake,\nthen I have to get used to it first and that takes time.\nAnd that's why there are difficulties, really big difficulties,\nwhen AI's hallucinate or when they, for example, have a bias that is then reproduced.\nFor example, that SQL injection is no longer a topic\nor that there are only SQL injections and the newer things that you should actually pay attention to,\nlike LLM injections or prompt injections,\nthat you don't even know that as a mistake,\nbecause it's just not in the training data yet,\nbecause it's too new.\nAnd of course also restricted AI systems,\nespecially Gemini, by the way, is a candidate there\nthat doesn't allow me security things, for example.\nIt's a big, big problem if I get stuck at some point.\nBut what I also noticed,\nand that is probably my main point of criticism,\nwhy you shouldn't just trust every output\nand shouldn't just let the AI do it,\ncritical thinking becomes weaker.\nAnd that means in cybersecurity,\nyou lose the context,\nyou lose this hacker thinking,\nwhich you just need.\nHow do attackers really proceed?\nHow could you as an attacker perhaps otherwise still proceed?\nYou often work out exactly this thought process\nvery, very long and very laborious in a certain way.\nHow could you actually attack a system at all?\nIf I don't do it myself and only take over from AI,\nwhat should actually be done or is done at all,\nthen I lose it.\nAnd that's why it's not easy, of course,\nto trust AI's output without further ado.\nBut you also have to say,\nall the people who tell you,\ndon't use AI at all, that's just shit\nand do it yourself,\nthey're also on the side.\nYou have to accept that very clearly\nand yes, you can't ignore that.\nSo why should you use AI at all?\nWell, quite simply because I personally\ncurrently have a factor in me,\na factor three or even higher\nin speed boost can determine.\nI always measure that again and again\nso that I have such a rough estimate.\nAnd at the beginning it was extreme,\nwhich I was more likely to do.\nIn the meantime, it has settled in quite well,\nbut the quality has improved again.\nAnd I have to say, with a factor three\nyou can't just say, well, then I'll leave it.\nFactor three means that you can work three times faster,\nyou can learn topics three times faster.\nWhy shouldn't you use that?\nPlus it is a very, very competent mentor\nwho is available at any time,\nwho can ask questions at any time,\nwho can sometimes be very, very niche.\nAnd it also has the competence\nin other areas of expertise,\nwhich is also incredibly helpful\nfor us as IT and cybersecurity experts.\nI'll give you a few examples.\nRight, I don't really know Jura.\nI have my reference points\nand of course I also have lawyers\nthat I regularly ask.\nIf it really comes into use.\nBut to get a first assessment,\nI have very, very often\nquestioned my legal insurance\nor questioned friends of lawyers.\nAnd let's put it this way,\nthey are not always available\nand of course they also make mistakes.\nAnd C, well, I have to pay them\nand sometimes it can be quite expensive.\nSo if I get a first assessment\nwhere it is immediately clear that it is illegal,\nit does not do that or where it is immediately clear\nthat it is not a problem, it definitely does that.\nWhy shouldn't I use that\nbefore I actually ask the real lawyer?\nI need as a self-employed person,\nwhich you might also want to be,\nI definitely need to know about finance,\nabout BWL, about tax law.\nAnd AI is actually very, very good there\nbefore I ask my tax advisor.\nI now ask her mainly in a finished letter.\nThat's my plan.\nThat's why I think it works, yes or no.\nThat's the idea.\nAnd that makes her very, very much cheaper for me\nbecause I don't spend so much of her time anymore.\nAnd at the same time I can also\nget an assessment myself very, very quickly.\nAnd psychology is of course also extremely important.\nHow would you do social engineering?\nThere are a lot of psychological components\nthat we, to be honest,\nare not trained for.\nUsing AI for that is incredibly helpful.\nAnd you really shouldn't miss out on that.\nSometimes we are also on the front page.\nSometimes this is exactly this.\nWhat else could you do?\nThe crucial point.\nMaybe I, as a human being,\ndon't recognize all the possibilities\nthat one could find for an application\nto carry out attacks.\nAnd there is a brainstorming with an AI system\npartly extremely helpful.\nSo it is not an option\nto completely ignore it.\nBut it is also not an option\nto completely trust the statements.\nAnd here we want to dive in more closely.\nWhat exactly do we want to do?\nHow exactly could we use it?\nAnd a large part of it is of course exercise.\nAnd here I want to come to our today's partner,\nnamely the Potsdam Cyber Games.\nThese are CTF challenges\nin a large event,\nthe Potsdam Cyber Games,\nfor beginners to professionals.\nThat is, they are specially designed\nfor beginners.\nBut above all, it should be clarified\nwhat is happening.\nSo what could be?\nWhat if?\nWhat could happen\nif an application is really unsafe?\nWhat could really happen\nif you have a certain weakness?\nAnd you get hacking challenges for that.\nChallenges that are even linked\nto teaching material.\nAnd there are also additional extra challenges\nthat convey the whole thing to you\nas building knowledge, so to speak.\nThere are 20 big challenges, so to speak,\nin various categories,\ni.e. web, forensic, network,\ncrypto, i.e. cryptography,\nhardware and of course porn.\nThese are the very classic\nHey, how do I get root rights challenges?\nAnd exactly these are organized\nso that you can collect exercises,\nso that you can, so to speak,\nand I think that's such a great concept.\nThat's why I want to look at it\na little more in detail here.\nThese challenges are not organized\nby anyone, but by an internationally\nsuccessful CTF team\nfrom the Hasso Plattner Institute.\nThis is an IT institute\nat the university in Potsdam.\nThat means we have experienced\nplayers and players\nwho in the end provide you\nwith their own challenges\nand a whole platform\nthat gives you\nas simple and easy access as possible,\nbut still shows real challenges,\nso that you can do\nthe perfect hands-on experience.\nAnd there are even prizes\nto win for you.\nThat means you can actually\ncome to Potsdam for a week\nin a final\nand get a workshop there\nplus a conference\nwhere you can do live challenges\nand then also with hands-on experience\nsimply connect with others on the spot, exchange\nand learn from people\nwho have a lot of experience in this area\nand simply want to pass on\nthis experience to you.\nYou also get the opportunity\nto get visits from exciting companies\nwho deal with cybersecurity.\nPlus, as I said, this conference\nfor national cybersecurity.\nThis is one of the largest\ncybersecurity conferences in Germany\nafter the Munich Security Conference\nwith live reporting from television\nwith senior representatives\nfrom business,\nfrom science,\nand even from the authorities.\nThat means BSI and BND are also represented.\nYes, you can actually\nparticipate in this conference.\nYou really get free tickets.\nThese are the prizes you can win here,\nwhich I find very, very cool, I have to say.\nBasically, these entire\nPotsdam Cyber Games\nare designed so that you can\ntake a step in cybersecurity,\nno matter what level you are at,\nwhether you are a beginner\nor advanced.\nYou can start directly,\nso actually not in a week or something.\nThe Potsdam Cyber Games are open,\nthat means you can register directly\nand you have time until the end of April\nto solve these challenges\nand to farm points.\nI'll put the link in the description below.\nYou have nothing to lose,\nbut you can definitely\nsimply win a very, very exciting career\nif you want to.\nAnd I can just recommend it.\nThere is literally nothing negative for you.\nJust to gather experience\nis incredibly valuable.\nI want to show you\na challenge\nand from that we go a little bit\ninto how we can use AI.\nThat means I'll leave\nadvertising here at the top,\nbecause the challenge is a challenge\nthat you will see at the Potsdam Cyber Games.\nThat means it is not from me,\nbut from HPI.\nBut yes, it's about the challenge.\nSo, the following happens.\nThis is the challenge Hex Chat.\nThis is an Android security challenge\nand it's actually\na encrypted chat.\nThat means you have a chat application\nand it has a password in it.\nSo,\nyou have to think about\nhow can I crack it?\nWhat could an attacker do?\nIt's about developing\nthis awareness.\nHow does a potential attacker proceed?\nAnd if you do this at the Potsdam Cyber Games,\nthen you can also get\na virtual machine\nand that's the one with the plane.\nIt's just called an airplane.\nBy the way, it's a cool name.\nAnd then you can do the Hex Chat Challenge directly.\nAnd it looks like this.\nThat means you have an Android application\nthat you get and you have to\nsomehow get access to it.\nAn Android application is basically\nwell, theoretically\nmost people download it via the Play Store,\nbut basically these are APK files.\nSo, APK\nis the file ending.\nIt's just a normal application.\nJust like .exe is a Windows application,\nAPK are applications for your smartphone\nthat it can execute.\nAnd that's exactly the kind of application\nyou want to crack.\nNow you have to think about\nhow do I get to this password?\nHow can I read the chat?\nWhat potential weaknesses is there?\nI mean, is it a challenge?\nThat means there will definitely be weaknesses.\nOf course, you can make sure of that.\nAnd these are your learnings\nthat you have learned.\nSo, how do you get to a password?\nWell, I can get it in several ways.\nIt could be in the app,\nit could be hard-coded,\nit could be transmitted via the network,\nor it could be on a server\nthat you have to attack.\nThese are several possibilities.\nIn the end, everything that the app communicates\ncan contain this password.\nAnd then you have to investigate\nwhere is it really?\nHow do I really get there?\nWhat are the various disciplines\nto carry out?\nBecause we are here in a practical use case.\nThat means it really has to be considered\nwhat can I actually do?\nAnd then you could suddenly start\nto use AI.\nFirst of all, of course, you need to think about it.\nI want to look in the app.\nWhat tools can I use\nto search for a password in such an app?\nWhere, for example,\nare strings\nin an Android application?\nHow exactly does, for example,\na password exist?\nWhich hash is this hash\non the right side?\nThese are all questions\nthat you can ask an AI.\nAnd you get an explanation.\nYou have to read through this explanation.\nThat means the AI does not directly solve\nthe task for you,\nbut it gives you knowledge\nwhat you can actually google in the end.\nBy the way, I had this hash over there\nthat I just generated for this video\nbecause I wanted to give you an example.\nAnd some of you may have even recognized it.\nClaude, if you ask\nexactly this question here,\neven wrongly written,\nthen the answer is\nthis is the MD5 hash for password\nwhich I found incredibly fascinating\nthat it was simply really learned.\nI asked nothing else.\nYou can try it out.\nI just found it mind-blowing\nthat it knows the hash value\njust by accident.\nBut yes, that should of course only show\nwhat you could do with everything.\nHow do you actually convert hashes back?\nHow do hashes actually work?\nAnd so on and so forth.\nOf course, you could talk about it.\nThat means you can ask questions\nabout the things.\nYou can learn about things\nthat you are currently trying to understand.\nAbout what you are currently investigating.\nHow does network traffic actually work in an app?\nHow can I perhaps view it?\nWhat tools can I use to do this?\nAnd why are passwords\nnot saved in Klartext, but as a hash?\nWhy isn't that always safe?\nWhat are API headers?\nWhat is JSON?\nHow can I handle HTTPS in an app?\nAnd what is certificate pinning?\nAnd yes, they all come together.\nAnd the trick here is\nthat these questions often\nare based on each other.\nThat means you can ask questions.\nAnd that is something that you could\noften not do with a classic Google search.\nThat means I can actually take\na lot with me here.\nI can very easily just say\nI'm going to dive deeper\ninto the respective matter.\nI'm going to dive deeper into each area\nas deep as I need\nand as long as I've solved it.\nAnd I'm gaining a lot of experience\nand everything I don't know\nI can learn.\nAnd this I can learn\nI can learn that with an AI.\nThey can explain it to me,\nteach me, be my mentor.\nAnd so it's really\nbasically like modern Googling\nfor explanations.\nIt's not a thing that gives you the solution.\nIt's like you're doing a CTF challenge\nand going through a work\na walkthrough.\nAnd that's exactly what we don't want to do.\nWe want to get to the solution ourselves.\nIn other words, it's basically\nexactly the same as what I did\nfor my studies, which is really Googling.\nOnly that you can actually go here\nand you don't have to Google\nbecause Google just doesn't work well anymore\nand didn't work well back then either,\nbut is packed full with posts\nthat are just insanely bad.\nYou can only scratch the surface\nbecause most of the traffic can be farmed.\nYou can actually go in here\nvery, very deep\nand then really get into the functionalities,\ninto the algorithms themselves.\nThe AI systems\ncan do that in most cases.\nYes, sometimes there is false information,\nbut if you ask such questions here,\nthen you often don't scratch so deep\nthat the AI systems can't do it anymore.\nFrom experience, I can actually\nat least give you that for the moment.\nOf course, something like this can always be worse\nbut from experience it gets better\nand not worse anymore.\nAnd if you need an actual recommendation\nfor April 2025,\nwhat do we have here?\nThen I would recommend you\ncurrently, because you probably don't want to pay for it,\nGemini 2.5 Pro.\nYes, no joke, I'll make another video\nabout it on another channel.\nOr rather, if you see this video here,\nit should already be there.\nI'll link it to you in the end card.\nIf you want to pay for it,\nI would recommend you to use\nExtended Thinking.\nThis is also a very, very good model\nespecially for IT users.\nThat was what I showed earlier.\nOkay, so basically\nall you have to do is get the information\nand continue with it.\nBut what does it look like\nif you want to go beyond these basic questions?\nI mean, that was already\na beginner's challenge.\nThese are questions that you probably have internalized at some point\nand where you know\nwhat is actually happening.\nYou really have to go into\nthe individual areas of expertise.\nAnd yes, cybersecurity has many areas of expertise.\nI have picked out the biggest,\nat least I would say so,\nand try to give you an overview\nof how to get\nout of the beginner level\nand maybe ask the questions as a junior.\nAnd here it is very, very important\nthat you want to gain experience.\nFirst of all, let's start with Security Operations.\nSecurity Operations,\nyou could ask something like\nexplain to me this log of, for example,\nan event.\nPay attention to the fact that\nyou do not give sensitive data\nto these AI systems\nas long as they are not running locally\nor maybe hosted by the company.\nChatGPT, Gemini, Cloud,\nthey all use your data to learn.\nYou transfer data to a third party\nand that means you have to deal with\ndata protection beforehand.\nThat's why the exclamation mark here,\nthis attention.\nBut you could also say, for example,\na certain attack.\nAnd, for example,\nwork with deep research,\nthat really writes you down\na real analysis\nand then, of course,\nread through the individual sources.\nWhat you should not do,\nfor example,\nis to post a lot of logs\nin an AI and then say,\nhey, analyze,\nwere we attacked?\nThat is obviously wrong,\nand would execute it.\nSo, the problem at this point\nis simply that you\ndo not learn anything\nwith this orange written command.\nYou do not know why we were attacked.\nWhat do these logs tell me?\nWhat are IOCs? What are IOAs?\nYou do not know all that.\nAnd that's exactly what you need.\nTo explain the terminology,\nwhat is an IOA,\nwhat is an IOC,\nwhat is actually a log,\nwhat is an event, for example,\nthat would be the area\nthat you have already learned in the beginner part.\nLet's look at the next thing.\nThreat intelligence.\nThreat intelligence is about\nanalyzing threats\nand identifying\nhow strong or how bad the whole thing is,\nwhat is actually happening\nand what can I do about it.\nFor example, you could ask,\nhey, this is a CVE,\na common CVE,\nwith a certain number.\nAnd, for example,\nwith a code fragment.\nThere is often a piece of code\nwhere you can just see\nwhat the whole thing is doing.\nYou can ask, what is the impact\nand what exactly is it doing?\nAnd maybe you even have a possibility\nto ask more questions.\nThe point is, however,\nthat you still have to know,\nokay, is this CVE relevant to me at all?\nAnd the problem with CVEs\nis often that they fall into\na completely different area of expertise\nwhen you, let's say,\nin an application\nthat you don't know at all,\nhave a CVE,\nbut you have to investigate it.\nThen you really have to work your way in there.\nAnd with something like that,\nof course, an AI system can help extremely.\nThen of course there is also the possibility\nto ask for mitigation.\nThat is, how can I actually prevent\nthat, for example,\na cross-site scripting weak point\ncan be executed on me?\nWhat would you normally do there?\nI wouldn't be asking\nif there is XSS,\nbut I would ask,\nhow do you normally do it?\nHow can I deal with it?\nAnd what I wouldn't do is to ask,\nis this IP address evil?\nOr write a complete report\nabout any APT,\ni.e. about an Advanced Persistent Threat Group.\nThat, in turn, is something\nthat you would of course\nlook for yourself\nand puzzle the information yourself.\nYou first have to learn,\nwhat is relevant to me\nin such a report?\nYou wouldn't learn that\nif you just let yourself do it.\nSo, the next area would be\nreverse engineering.\nAn example of this,\nreverse engineering means\nthat you get from a malware\nthat you might get as a fragment\nor as code or also as an executable file\nthat you want to analyze them\nand want to know, okay,\nwhere do my attackers come from?\nOr what exactly does this malware do?\nAnd then you could say,\nhey, I have this\nassembler fragment here,\nor these one, two, three lines of instructions,\nwhat exactly are they doing?\nExplain to me exactly\nwhat they are doing there.\nOr, what I can actually\nhighly recommend,\nwhat does this Windows API function\nactually do here?\nIn exactly such fragments\nit often says,\nhey, this is a call to this function\nbecause it is a Windows internal function.\nThat means you are attacking\nthe Windows interface.\nAnd well, you have to know\nthese internal interfaces.\nBut there are so many of them\nthat you can really know them all\nin the rarest of cases\nand you don't have to know them.\nNow you would just go here\nand say, hey, I'm googling for them.\nBut I can also ask exactly\nAI questions and ask, hey,\nwhat exactly does this API function do there?\nAnd well, what does that mean for me?\nWhat it shouldn't do\nis simply to post this\nassembler package in there\nand ask, hey, what exactly does\nthis code fragment do there?\nOr, hey, here is assembler\nor here is malware,\nanalyze it, where exactly\nis the C2 address?\nThat's exactly what you shouldn't do\nbecause then you won't learn anything\nbut the AI will let you do\nall the work.\nYou can see where this leads.\nNow let's look at other areas\nlike app security.\nFor example, why is this\ncode area accessible?\nThere should be as little code as possible\nand not just a\nfind all weaknesses in my codebase\nbecause that makes no sense\nbut you really want to analyze\nindividual lines.\nWhy exactly?\nHow exactly would you attack\nexactly these lines of code?\nSo really understand the nitty gritty\nbefore you let it be done\nor how would you\nperform a parameterized\nSQL query\nin, let's say, PHP\nor even Python\nso that it is no longer accessible?\nHow would that look like?\nAnd please check that\nbecause there are occasionally\nstill errors.\nApp security, in case you didn't notice\nmeans of course the application security\nSorry, I didn't say that just now.\nThen of course we also have the\nCloud security area, that is, the security\narea.\nBut you also have to know\nwhat is the difference\nbetween AWS security groups\nor network ACLs.\nYou have to know that first\nbefore you can set them up, of course.\nBut I don't let them be set up\nby an AI completely,\nbut what is the difference\nbetween the two.\nAnd I can ask, for example,\nwhat makes a certain property\nin a certain area\nthat I should set up at, let's say, AWS\nis a complete configuration\nor any terraform script\nor any administrative software\ncompletely to be written\nif you don't even know\nwith the respective tool.\nThat is, please learn terraform first\nbefore you can write terraform scripts.\nLogical.\nAnd then the last area\nthere is of course also GRC,\ni.e. Government Risk and Compliance.\nIt's about how do I actually manage\nmy application, that there is really\nnothing coming out. How am I actually\nconform or what\nwhat risks do I actually have?\nAnd here it is now\nfor example about\nwhat is actually a safe date?\nSo PII stands for\nPersonal Identifiable\nPersonal\nPersonal Identifiable\nInformation, so\nperson-related data, basically.\nIs an IP address, for example,\nis that missing under PII?\nIf so, how can I protect them?\nHow can I, for example, from my logs\nremove? If no,\nthen I can continue to use them.\nWhere exactly is the limit there?\nOr also do a deep research.\nWho is actually really affected by NIST 2?\nAnd then of course the deep research\nmainly use to find these\nresources and then\nread through. Something like a better\nGoogle search. As I said,\nagain, what you shouldn't do\nis for example a complete\nrisk assessment for any system\nor your security policy\nto write. You should do that\ndo it yourself. And if you\ngo through all this, then\nyou still have\nthe main work and that where you really\nthen you have to do that\nyou still have it with you. You use\nbasically also here\nAI systems as, I say,\ndeeper knowledge\ndatabases, which you\ntechnically also can google.\nWhich I have googled a lot.\nBut that's how it goes very, very\nmuch faster. And in a fair way, who\nhas not googled anything yet, the\nwe say who his source is.\nBut that means, if we go back\nagain to the observations, to the\nProblems I mentioned earlier.\nThe use of AI, that we do not understand\nwhat is actually happening there and\ncan therefore ask worse questions.\nThat we stay stuck, as soon as the AI\nstays stuck, because we can't do it ourselves\nand that we don't think the hacker\nno longer learn. Basically, we build\nhere our own understanding.\nWe learn these things, yes. We want\nunderstand how it works in the back\nand we use the AI for that. We ask the\nAI not directly for a solution.\nWe do not stay stuck,\nwhen the AI suddenly has no idea\nbut we know fundamentally\nwhat we do and we use the AI\nonly as a kind of google replacement,\nwhich works better and works faster.\nAnd above all, we lead\nthrough. That means we learn\nalso how to think as an attacker.\nWe train ourselves, yes.\nAnd we don't let the AI just do it.\nThat means the three big problems\nPoints, they actually fall here\naway. While at the same time\nthe big\npositive points, they remain\nstill available.\nWe still have the opportunity\nwith AI to build up our knowledge\nmuch faster, much deeper\nto go into a deep dive without\nreally much with superficial posts\nor any SEO-optimized\narticles to fight.\nWe have the opportunity here\nto have a very competent mentor and\nwe take advantage of the fact that you\nour blind spots can also then\ncan increase. That we in the end\njust go in there more\nand can just learn there\nwhat we might not have known before.\nAnd all that\nwe do of course\ngraded.\nYou will as a beginner\nand as a junior a completely\ndifferent use of AI with you\nsee how you have it as a senior.\nThat means if you are learning\nthen you still learn for\nyou. You can go into every topic\njust so deep dive. You have every time\nsomeone who advises you, who you with\ngives you knowledge and who\nsupports you in deep diving.\nYou have the opportunity to brainstorm,\nyou have the opportunity to learn a lot of context\nhow that works, why\nthat happens and what is actually\nbehind it. And you get explanations\nto really every topic\nthat you want to know.\nAt the same time you also have something\nthat basically takes away from you\nthings that you in your expert area\ndo not necessarily need.\nSo a kind of first assessment\nfor legal topics, for BWL, for\npsychology, I don't really need\na degree for that either.\nAs I said, I don't have a degree in BWL\nI don't have a degree in psychology\nor in law. And yet I have to\nwork with these areas of expertise\nof course. I have my experts I can ask\nbut when you start\nthen you often lack the money for something like that\nand for that you can also use AI.\nSo, as a junior, on the other hand\nthat is the second area here,\nyou mainly want to gain experience.\nThat means you want to learn, okay,\nhow does it really feel in practice?\nYou just want to do it.\nAnd here you have\nthe opportunity to use AI as a mentor\nas a personal coach\nwho really supports you very specifically\nfor this application case.\nAnd there is no such thing as AI\nat all. You might have been able to Google\nbefore and maybe you were lucky enough\nto find something on Stack Overflow\nthat fit you pretty well\nbut not exactly on point.\nThat means you can ask very specifically\nfor individual logs, very specifically\nfor individual questions, where you then\nget your own explanation\nto exactly what you need.\nBut you still have to\ngo through everything yourself.\nThis way you get the routine, the practice\nand of course the experience\nthat is so incredibly valuable.\nAnd with AI you have a kind of\nwell, when programming you say per programmer\nI called it per analyst\nwho will help you if you ever get stuck.\nYou don't have to look at a complete\nwalkthrough of any CTF.\nI've done a lot of that\nand that's how I was able to\nspoil it completely.\nIt's logical, isn't it?\nBut you can now just say\nattention, please don't spoil me.\nAI can really say that.\nI highly recommend it.\nAnd then after one step,\nwhat you might still be able to do,\nask. Not directly after a full spoiler\nbut you want to collect the experience.\nAnd if you have this experience,\nthen at some point you will be\nin the lower area.\nAnd then you are on your way\nto the next level.\nAnd there it is then so\nand that is fair enough\nnot different from what we have done so far\nthat others are allowed to do AI\nand then of course\nhave to deal with this AI system.\nAnd if we look into practice\nand are very, very honest\nthen seniors use very, very often\nor also seniors in the sense of\nproject management use very, very often\nforeign workers who are cheaper.\nNow, in the meantime, they let\nthe AI do the work\nand then you look, okay,\nhow good is that?\nWhat do I really have to improve on?\nAnd you have to work over it anyway\nbecause there are errors in it.\nBut that's no different\nif I let someone else do it\nwho may not have any experience\nin the field. That is completely\nremained the same.\nAs a senior, you still have to\nwork with the results\nand pack them together, process them,\nwork them over again in a meaningful way\nWhat can't I give to AI?\nWhat can't I just give to information\nout there because, for example,\nthere is a data protection problem?\nWhat can AI maybe not do well\nand what can it actually do how well exactly?\nThat means, if I let something do\nthen I have to review it again\nand basically that is\nno new workflow\nfor a senior.\nAs I said, that's what they did for years.\nThat means, nothing changes,\nonly that the tools in the end\nbecome much more efficient\nand much more powerful\nand can better support you\nin learning, in gathering experience\nand also in doing\nwhere you only have to read over it.\nBut it depends on\nhow high your experience is\nand how much you can\nassess yourself.\nWell, that would be\nmy take\non how you can best use AI\nto work in your field.\nI am 100% sure\nthat AI is the best\nand that there are some people\nwho have a completely different opinion\nabout it and also\nhave a completely different\nexperience with it.\nSo if you are someone like that,\nfeel free to write in the comments\nwhat exactly your experience is with it.\nI would highly recommend\nto build up a knowledge treasure\nin your field\nand a knowledge treasure\nregarding AI.\nNobody will take it away from you\nuntil you have it.\nAt this point,\na short recommendation from me.\nAs you may know,\nwe have the MorphReader\non PlayStore and AppStore.\nI am currently working on the third version\nwhich will be completely reworked.\nYou can still use\nthe current version very well.\nIt is also finished so far.\nThere are only a few optimizations\nthat I would like to do.\nThat's why I changed the architecture\nand there will be a separate video about that.\nSo you can see\nwhat I am currently working on\nand maybe also new videos.\nSo feel free to check it out.\nYou can also find me on Patreon\nif you want to use the RSS feed\nin another reader,\nthe raw RSS feed.\nSo thank you very much\nfor being with me.\nI am very excited to hear from you.\nSee you next time in cyberspace.\nCiao!",
  "transcript_chars": 32050,
  "ingested_at": "2026-05-15T10:34:21.964193+00:00",
  "source": "channel",
  "yt_meta": {
    "view_count": 12722,
    "like_count": 527,
    "channel_id": "UCLGY6_j7kZfA1dmmjR1J_7w",
    "categories": [
      "Education"
    ],
    "tags": []
  }
}