Fayaz Ahmed’s Post

Ever wondered what goes on behind the scenes of your favorite web browser? 🤔 Dive into the world of Web Browser Engineering and discover how AI helps streamline and secure your online adventures. From rendering pages to protecting your data, it's like having a digital superhero in your pocket! 🦸♂️ Check it out and imagine the possibilities of AI at your fingertips. Read more here: https://lnkd.in/ge--B3An. What's your favorite browser feature? Let's discuss!

AI is taking over your web browser

The latest Google Pixel 8a leaks hint at its design, software updates, and AI features https://lnkd.in/gvmJ3iKy

I have given Googles BERT and T5 models the same PDFs as I gave to Chat GPT to compare them, and I got significantly different results. Not sure if I have missed something on the documentation on how to implement them but the performance of T5 and BERT was very bad for this task. Have a look at the 2 summaries they generated. I have also tested questioning the text and the performance was the similar. Google's Gemini is a more fair comparison which I will be running next! Full video on how I created the 2 apps here: https://lnkd.in/duRpF_s2

Chrome is adding a tiny GPT model right inside your browser. That's massive! AI apps at zero cost, with no server infrastructure? 🤯 I can think of so many use cases: 🧪 Generating landing page copy on the fly, ⚡️ Smart autocomplete, autocorrect, search & suggestions, 🤖 Tiny built-in chatbot assistants, 📈 Next-level personalisation Besides, everything happens without an internet connection, so it's: 🔐 100% private, as no data is shared, and 🕒 Lightning-friggin-fast! Do you think this is the next thing for the web? Credits & original tweet by: Vercel & Guillermo Rauch

#Google_lens #AI #Gemini AI battle is getting intense. Google is bringing lens to chrome web browsers also much like circle to search feature in mobiles. What it really means for end user 👉 Search and ask questions about anything you see on a webpage. 👉 Select text, images, or objects to get more information. 👉 Get answers and results directly within your Chrome tab. In ver simple language, Instead of just looking at things on your screen and then google them, you can ask Google about them on the same tab in a click! 👉See a cool product? Lens can find it for you and tell you where to buy it. 👉Don't understand a word? Lens can translate it right there on the page. 👉Found a funny meme? Lens can find similar memes or images. 👉Need help with a math problem? Lens can solve it for you. 👉Basically, it's like having a helpful assistant right there on your computer screen, ready to answer your questions about anything you see! #Productmanagement Post link in the comments

Enabling Full Local LLM Execution within Browsers via Direct GPU/CPU Access

I developed a Computer Vision application utilizing Azure's Computer Vision API. The app is designed to analyze images, extract insights, and recognize objects, text, and scenes with high accuracy. Key features of the application include Image Analysis: The app can analyze images and detect various visual elements such as objects, people, and activities, providing detailed descriptions of the content. Face Detection: The app is capable of detecting human faces in images, including attributes such as age, emotion, and gender.

So the #MediaPipe / Core ML team at Google launched a new #LLM API for Google's Gemma model today for #WebAI! Was pretty cool, so I whipped up a little demo tonight at home that goes beyond just talking to it like chat bot. Instead I wanted to provide utility to folk browsing web pages to see what you could do with just a 2.5B parameter LLM to boost your browsing experience - imagine this was a Chrome extension! Capture is in real time (#WebGPU is so fast!), on an old PC with an NVIDIA 1070 GPU - running entirely in the web browser entirely locally in Chrome - no server calls after the page load - so yep your privacy is preserved, and yep it could work offline. Want to try it yourself? Let me know in the comments and I'll look into getting a demo put up on CodePen once I figure out the logistics of hosting the model somewhere. Theoretically one could cache the model in local storage so you don't have to wait for any download on the next page load - but I'll leave that as an exercise for you. #WebML is rocking it in 2024 - it's time to update your Web Dev skillset with a splash of Web AI in JavaScript! Obviously this is the default unmodified LLM, if you were to fine tune or distill you could get even better results for select tasks you wanted to perform - this is just a demo and was made in my own personal spare time. Learn more about how you can try this model yourself here: https://lnkd.in/g-QVQrrF

As a consumer, I welcome this transition to more conversational and consultative version of search. But for marketers, what does this mean for having our brands found by consumers? The authors provide 3 ways for marketers to optimize search results for LLM apps, and importantly, they outline a new job taking shape within marketing orgs.