The 3 Scaling Laws of AI: From Training More to Thinking More

Hello, I'm Rijul, and I'm building LiveReview — a blast-radius aware AI code review built for your business-critical systems. Star us to help devs discover the project, give it a try, and share your feedback to help improve the product. We see the pace of AI development speeding up. One way to view this is through the 3 scaling laws of AI. It originally started with one scaling law, but as AI has advanced, it has grown into 3 different ways of scaling AI capabilities. Pre-training Scaling This is the scaling involved in the pre-training of a model. To help understand it better, imagine a student preparing for a huge general knowledge exam. If they read 10 books, they might do okay. But if we scale that to 100 books, they will do much better. So, the more they study, the more their score climbs. Now, coming to a real example, GPT-2, which was released in 2019, had 1.5 billion parameters and was trained on a modest slice of internet text. GPT-4-class models were trained on trillions of tokens. The difference in knowledge between them is a direct result of pre-training scaling. Post-training Scaling Now imagine that same well-read student takes practice tests, and a tutor gives them feedback: "That answer was too long." "You should have shown your reasoning." "That tone was rude." The student's knowledge doesn't grow, but their ability to give the answer the grader actually wants improves enormously. Coming to a real example, the raw pretrained GPT-3 model had issues with going off-topic. But after reinforcement learning with human feedback, it became what we know as ChatGPT. The underlying knowledge is largely the same, but its ability to follow instructions and present answers in an ideal way has improved. Inference Time Scaling Picture 2 students solving a tricky math problem. One student instantly blurts out the first answer that comes to mind. But the second student works it out on scratch paper, checks their steps, tries different approaches, notices their previous mistakes, corrects them, and eventually reaches a solution. So, the second student gets the correct answer because they spent more time thinking, not because they studied more beforehand. Now, this is what we see with Agentic AI. We can see different effort levels, and the system can determine how much thinking goes into generating a response. So, a simple analogy to understand this is: Pre-training makes the student smarter, post-training makes the student better at giving good answers, and inference-time scaling makes the student take their time on hard questions. Wrapping Up It's interesting to see how AI is evolving and how fast-paced this evolution is. When these 3 scaling laws come together, it's even more impressive to see how AI capabilities are continuing to improve, with GPT-6 Astra coming up and the capabilities it is offering. Your team's attention is limited, and the deluge of AI-generated code is making it harder to keep production reliable and secure without slowing you down. I'm building LiveReview, a blast-radius aware AI code review built for your business-critical systems. Instead of presenting every diff with equal emphasis, LiveReview scores each change by blast radius — how far its impact reaches through your call graph — so you can focus attention where it actually matters. Spend code review effort where business risk is highest — not spread evenly across every diff. ⭐ Star it on GitHub: HexmosTech / LiveReview Blast-Radius Aware AI Code Review for Business-Critical Systems       LiveReview: Blast-Radius Aware AI Code Review for Business-Critical Systems LiveReview is an AI code reviewer that scores every hunk of a diff by blast radius: how far a change reaches through your call graph, how much persistent state it touches, and how well-tested it is. A 3-line change to a shared auth check can outrank a 300-line UI tweak. Your team's attention goes to the highest-risk code first, not spread evenly across every diff. blast-radius-demo.mp4 LiveReview's Blast Radius & Review Priority scoring, live in the diff viewer. The exact math, not a black box Visualize blast radius at a glance Every factor that feeds the score How does Blast Radius scoring work? (a more technical explanation) Here's the goal: A 3-line fix in a function used by 40 other files, that also writes to a database, should score high. A 300-line UI change in one file, fully covered by… View on GitHub Click below to try LiveReview with your codebase: