Dr. Sajid Farooq

This is how Large Language Models (like ChatGPT) work: GPT stands for “Generative Pre-Trained Transformer”. Let’s break down what this means. Generative: this means the model can write (or “generate”) new text. As opposed to simply analyzing or processing inputs, it can output totally new, unique text. Pre-Trained: when you hear about AI being trained on massive data sets, this is what they are talking about. In the pre-training phase, they give the model tons of data, which in the case of LLMs, it uses to learn the patterns of language as well as general knowledge. Transformer: this is the key that has made all modern LLMs and the recent explosion in AI possible. In 2017, a team at Google introduced transformers in a landmark paper called “Attention is all you Need.” Transformers allow the model to analyze every word in a sentence simultaneously in order to understand which ones are most important. (Previously language models tried to analyze words in sequence and wound up having memory issues). — Ok, so how does this all work? At the beginning they start with a shit-ton of data. They can’t just dump it all on the Large Language Model, it needs to be cleaned up and formatted properly. This takes a lot of time. Next, all the words need to be turned into numbers. Computers don’t actually understand words, so the words are broken up into “tokens” and assigned a unique number identifier, which the computer can use to reference that word. They then use these numbers to create an insane multi-dimensional map of the entire language. And when I say multi-dimensional, I’m not talking about 3D space like our puny human brains can comprehend, I’m talking about THOUSANDS of dimensions. They place these word-numbers (“vectors”) into a multi-thousand dimensional space map. This is essentially a meaning map that allows the computer to comprehend language. All existing words are placed somewhere on the map based on their meanings and their relations to each other. This is what it uses to predict words. The next phase of training is basically playing a prediction game where the model is given examples of sentences and asked to guess what the next word will be. If it guesses wrong, its “weights” are tweaked and then it tries again. This happens over and over again billions of times, until it gets really good at predicting words. Once it reaches a good place, its weights are fixed. There’s an additional layer of “fine-tuning” for personality, telling it what its role is, what style it should communicate in, and setting some parameters around how it should “behave”, but this is the basis of the mechanism it uses to generate text in realtime when you send it a prompt. — I’m curious if this makes sense or is still too technical? My goal is to make technical stuff like this more accessible, so would love any feedback if you’re feeling like: “what the hell are you talking about?”

3

Game studios run on creativity, not bottlenecks. P4 One makes collaboration effortless so artists can stay in flow and ship faster.

18

1 Comment

Kris Huang, ROG Committee Head and General Manager, Gaming Gear, has published a thoughtful piece, “From Insight to Innovation: ROG’s Design Thinking Approach to Gaming,” on why design thinking plays a decisive role in building gaming experiences that hold up in real life — not just on paper.   What stands out is the consistent player-first discipline behind ROG’s work: deep research across different gamer profiles, a structured loop from empathy to definition and testing, and the willingness to turn engineering constraints into meaningful design outcomes. The article also shares concrete examples — from addressing rigidity, acoustics, and thermals in the Zephyrus program, to rethinking feasibility and aesthetics in the Flow Z13, and building mouse ergonomics with hand-shape development informed by pro players.   If you work in product, design, or engineering, it’s a solid read on how insights become decisions — and decisions become better gameplay. 𝐑𝐞𝐚𝐝 𝐭𝐡𝐞 𝐟𝐮𝐥𝐥 𝐚𝐫𝐭𝐢𝐜𝐥𝐞 𝐡𝐞𝐫𝐞: https://lnkd.in/g_kUHMg5

8

1 Comment

After years of leading engineering teams and developing new-to-the-world products in optics, display, and instrumentation, I’ve learned that innovation thrives where collaboration lives. Magic Leap has shown how bringing together diverse expertise can turn complex challenges into real progress. This new direction of becoming an AR accelerator for technology partners is a powerful way to build on that experience. By combining our deep technical knowledge with the strengths of our partners, we can move the AR ecosystem forward faster and with greater impact. The future of augmented reality isn’t built in isolation. It’s built together. This article reflects that evolution and the shared vision driving it forward.

132

1 Comment

🚀 We’re officially developing the CollectiveOS mascot — and the process has been both technically fascinating and creatively rewarding. As we build out our animated and narrative ecosystem, we realized something important: a character that appears across video segments, explainers, branding, product UI, and future AR/VR experiences needs to be perfectly consistent from scene to scene. And that’s where today’s AI video models become a real challenge. Even with cutting-edge tools, characters tend to drift over time — 🔹 proportions shift 🔹 colors vary 🔹 facial details change 🔹 lighting and texture fluctuate 🔹 expressions aren’t always stable For a true mascot, that variability isn’t acceptable. So we’ve been testing multiple versions of our character — refining textures, glow levels, gill design, expression style, neon aesthetics, and proportion geometry. The goal is simple: lock in a design that’s instantly recognizable, emotionally resonant, and technically stable across repeatable 10–15 second video segments (the current sweet spot for generative video consistency). Once the design is finalized, we’ll be able to stitch short AI-generated sequences into coherent long-form storytelling — from product intros to narrative worldbuilding to educational explainers. The mascot’s name is Kibo, and he represents something we believe in deeply: ✨ hope, clarity, and the human side of advanced systems. We’re close — and we’re excited to share his final form with the community soon. #CollectiveOS #AIMascot #AIInnovation #CharacterDesign #GenerativeAI #Storytelling #TechBranding #DigitalCreativity #FutureOfAI #NeonDesign #Worldbuilding #EmergentSystems

1

Recent developments in new glasses-like form factors for XR headsets have got me ruminating on a long-standing immersive learning challenge: 𝐅𝐑𝐈𝐂𝐓𝐈𝐎𝐍 Will converging trends such as: 🤓 Lighter, less intrusive XR headsets, 🤖 Context aware AI, and 🔌Always-on XR 𝘧𝘪𝘯𝘢𝘭𝘭𝘺 signal the solution? I'm going to be exploring this in a series of posts over the next few weeks, so stay tuned! #LearningTechnology #XR #EdTech #WorkplaceLearning

12

3 Comments

It's research like this that makes Siggraph so special, scientists and researchers actually moving the technology of art forward. When HDR becomes more ubiquitous, it'll be because of research like this. This is where technology like Gaussian splats, stereo imagery, rendering engines get born and gain momentum.

3

A lot of tech press this week picked up on rumors about Apple releasing an AI “Pin” - a wearable with cameras, speakers, and microphones (and potentially other sensors) positioned as an “everyday AI companion.” Versions of this idea have been tried before (see the Humane AI Pin), but if anyone could drive mainstream adoption of a device like this, it would be Apple. The more interesting question, though, is not the device itself, but the interface that will ultimately allow AI tools to meaningfully integrate into our everyday lives. We already have access to powerful AI on our phones, but even the small act of pulling a device out of your pocket creates friction - a shift in attention and intent that keeps AI from feeling truly ambient. At CES this year, there were noticeably more AR glasses than in years past, suggesting momentum toward interfaces that surface information in a more natural, glanceable way. The technology to deliver contextually relevant information - and to do so in a way that is both socially acceptable and design-forward - feels increasingly close. Assuming similar research is underway at Apple, it will be fascinating to see how they apply their strengths in integration and industrial design to remove friction rather than add another screen. I’m excited to see where this goes, especially if the result is technology that genuinely helps in the background, without becoming yet another distraction from the world right in front of us.

20