https://virtueofvague.com/tags/AIFundamentals/Recent content in AIFundamentals on Virtue Of VagueHugo -- gohugo.ioenprakashpayyanagoudar@gmail.com (Virtue of Vague)prakashpayyanagoudar@gmail.com (Virtue of Vague)© 2026 Virtue of VagueWed, 24 Jun 2026 10:00:00 +0530https://virtueofvague.com/posts/ai-series-11/Wed, 24 Jun 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-11/ AI Series · post 11 of 12 series index → from noise to meaning — the quiet revolution in AI # every AI generated image you’ve seen started as pure noise. not a rough sketch. not a blurry draft. literal random noise. static. and then, step by step, something meaningful emerged from it. that’s diffusion. and it’s behind deepfakes, synthetic media, AI generated phishing assets, and some of the most realistic fake content circulating right now.https://virtueofvague.com/posts/ai-series-10/Wed, 17 Jun 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-10/ AI Series · post 10 of 12 series index → the AI everyone is talking about — how does it actually work # you’ve used it. you’ve been impressed by it. you probably don’t know how it works. that’s fine. most people don’t. but as a security professional — understanding what’s under the hood matters. because attackers already do. generative AI — the concept # previous posts covered AI that classifies, detects, predicts. generative AI does something different — it creates.https://virtueofvague.com/posts/ai-series-9/Wed, 10 Jun 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-9/ AI Series · post 9 of 12 series index → how AI sees images and reads sequences — two tools, one post # not all data looks the same. images are grids. logs are sequences. deep learning has a different tool for each. feed an image into a standard neural network — it loses all spatial information. feed a log sequence into one — it loses all temporal context. wrong tool, wrong result.https://virtueofvague.com/posts/ai-series-8/Wed, 03 Jun 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-8/ AI Series · post 8 of 12 series index → the brain analogy everyone uses — here’s what it actually means # deep learning is the reason AI got dramatically better. not magic. architecture. same data. same computers. different structure. suddenly image recognition works. speech recognition works. threat detection gets dramatically more accurate. the architecture changed everything. let’s look inside. the perceptron — where it started # smallest unit of a neural network. one decision maker.https://virtueofvague.com/posts/ai-series-7/Wed, 27 May 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-7/ AI Series · post 7 of 12 series index → learn by doing — how AI figures things out the hard way # how do you learn to drive? not from a textbook. from doing it, failing, adjusting. nobody handed you a perfect rulebook. you got in the car, made mistakes, got feedback, improved. over time your decisions got better because the feedback loop worked.https://virtueofvague.com/posts/ai-series-6/Wed, 20 May 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-6/ AI Series · post 6 of 12 series index → this is literally what your SIEM does — now let’s understand it # your SIEM doesn’t know what an attack looks like. it knows what normal looks like. everything else is an anomaly. that’s the entire foundation of modern threat detection. not rules. not signatures. a learned baseline of normal behaviour — and a flag when something deviates from it.https://virtueofvague.com/posts/ai-series-5/Wed, 13 May 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-5/ AI Series · post 5 of 12 series index → no labels, no problem — finding patterns in the chaos # some threats don’t come with a label. that’s where unsupervised learning lives. supervised learning needs examples. labelled data. known outcomes. but what about a zero day? a novel attack technique? behaviour you’ve never seen before and have no label for? you can’t train a model on what you don’t know exists. unsupervised learning handles exactly that.https://virtueofvague.com/posts/ai-series-4/Wed, 06 May 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-4/ AI Series · post 4 of 12 series index → drawing the line — how machines separate threats from noise # classification is easy when the data is clean. it never is. in the real world — malicious and benign behaviours overlap. a legitimate admin running powershell looks a lot like an attacker doing the same. a user downloading a large file looks a lot like data exfiltration. the line between threat and noise is rarely clean.https://virtueofvague.com/posts/ai-series-3/Wed, 29 Apr 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-3/ AI Series · post 3 of 12 series index → your SOC playbook is literally a decision tree # every escalation decision you make follows a pattern. machines do it too. is this alert high severity? yes — escalate. is the IP external? yes — check threat intel. is there lateral movement? yes — page the IR team. you’ve been running a decision tree in your head every single shift. the algorithm just formalises it.https://virtueofvague.com/posts/ai-series-2/Wed, 22 Apr 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-2/ AI Series · post 2 of 12 series index → teaching machines with examples — like training a junior analyst # here’s something you already do at work without realising it. every malware sample you label. every alert you close as true positive or false positive. every ticket you tag as phishing or legitimate — you’re creating labelled data. and labelled data is exactly how supervised learning works.https://virtueofvague.com/posts/ai-series-1/Wed, 15 Apr 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-1/ AI Series · post 1 of 12 series index → AI, ML, deep learning — same thing? not quite # three terms. endless confusion. let’s fix that. every vendor, every job posting, every threat intel report throws these around interchangeably. AI-powered detection. machine learning model. deep learning engine. sounds impressive. means nothing if you can’t tell them apart. so let’s untangle this once and for all.https://virtueofvague.com/posts/ai-series-0/Wed, 08 Apr 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-series-0/ AI Series · post 0 of 12 series index → why a soc analyst is learning AI (and why you should too) # so i was thinking about something. i spend most of my day staring at alerts. triaging. correlating. escalating. repeat. and somewhere between the 40th investigation and the third cup of chai, i realised — the tools i use every day are powered by technology i barely understand.https://virtueofvague.com/posts/ai-fundamentals/Wed, 01 Apr 2026 10:00:00 +0530prakashpayyanagoudar@gmail.com (Virtue of Vague)https://virtueofvague.com/posts/ai-fundamentals/a series for SOC analysts learning AI from the ground up. twelve posts. one concept per post. short reads. built for people who work on top of AI systems every day and want to understand what’s underneath. # post 0 why a soc analyst is learning AI (and why you should too) 1 AI, ML, deep learning — same thing? not quite 2 teaching machines with examples — like training a junior analyst 3 your SOC playbook is literally a decision tree 4 drawing the line — how machines separate threats from noise 5 no labels, no problem — finding patterns in the chaos 6 this is literally what your SIEM does — now let’s understand it 7 learn by doing — how AI figures things out the hard way 8 the brain analogy everyone uses — here’s what it actually means 9 how AI sees images and reads sequences — two tools, one post 10 the AI everyone is talking about — how does it actually work 11 from noise to meaning — the quiet revolution in AI