Marvin Liao’s Post

OpenAI’s announcement of ‘Advanced Voice Mode’ is only the beginning of a large boom in the space of voice interfaces and here’s why. In case you missed it, OpenAI announced yesterday that they will gradually release the Voice Mode they demoed in May, which is essentially a ‘very advanced Siri’. What makes this 'advanced Siri' possible is a paradigm shift in the model. For the first time, it is multimodal, meaning it doesn't have to transcribe speech to text, process it through the LLM, and then convert it back to speech. Instead, it processes speech input to speech output in the same way it would process text to text (LLM). AFAIK, OpenAI is the only company to release such a capability. It's groundbreaking as it reduces the speed of voice interactions to negligible/human speed and enables interruptions, among other features. Even without this paradigm shift, many builders were already creating companions (e.g., Character AI voice), personal assistants (e.g., Will.I.AM’s FYI app), and a ton more. The existence of these tools, despite the limitations of the previous modeling paradigm, indicates a strong demand for voice interfaces. And with OpenAI coming out with the paradigm shift model, while not accessbile publically, you can only asume that Meta (Llama), Mistral, Anthropic will soon also be releasing such a model, available publically. And when that happens, applications using voice interfaces (companions, assistants, games, etc.) will flourish. And as attention shifts to audio, advertisers will start paying more attention. And Wondercraft will power this attention.

Well why not?….All the cool kids are doing it🤓. The commoditization of AI has already begun. Curious how the AI Model game is going to play out a year from now. My .02 - it’s more about the data you have, how it’s organized, and the use-cases you’re solving for. See a lot people trying to buy AI like it’s a COTs app…it’s really more of a tool you BUILD with, not something you BUY off the shelf. #GenAi https://lnkd.in/g5Ddd-bC

Fascinating to see how Google AI Studio simplifies model development to deployment in the AI world. Using it as an SDK makes life easy for developers. Just a simple prompt to type in and develop the ML tools you need for production. Check out more details here: [Link to the article] #GeminiAPI  #AI #ML #GenAI Search for #GeminiAPI  #productionApplication #MachineLearning #DeepLearining #CNN #Regreesion #Classification #ObjectDetection #GoogleAIStudio #OpenAI

Just had a firsthand encounter with AI reshaping our daily interactions. Needed to reach out to Notion’s support. So, I quickly voiced my request using OpenAI’s Mac desktop app (highly recommend it, by the way). Seconds later, an AI MailBot replies, autonomously handling my inquiry. Basically, two AIs just chatted indirectly :) AI is becoming the new layer between us-automating the mundane so we can focus on what truly matters. As more people hop on the AI train, our workflows are bound to transform. And yes, the AI did forward my ticket to a human in the end. So, human connection isn’t dead just yet ;)

🤖 Why I Decided to Integrate AI into My "Word Game" Project? When I first started building my "Word Game", I used DynamoDB to store the words and Lambda functions with API Gateway for the game logic. It seemed like a great plan until I realized how slow it was to load thousands of words into DynamoDB. I knew I needed a better way if I wanted the game to grow. I looked into using a dictionary API,  but many of the good ones turned out to be quite expensive. I was feeling stuck until my brother Raymond Tran said, “Why not try AI?” At first, I wasn’t sure if AI would work for a "Word Game", but I decided to give it a try. That’s when I found OpenAI’s API, and it completely changed the game. It was easy to integrate, didn’t cost much, and made the game smarter and much more powerful. AI isn’t perfect—sometimes it doesn’t give me exactly what I want, and weaker models can make mistakes. Plus, there’s a small cost, but it’s not a big deal. Honestly, using AI has saved me a lot of time with managing the game logic. Instead of handling all the complicated rules and word processing myself, AI takes care of it quickly. The best part is how AI automates things, making everything run smoother and with fewer mistakes. It’s made the game smarter and allowed me to focus on making it more engaging for players. Even with a few challenges, the time it saves and how much better it makes the game are why I keep using AI. Play the game here: https://lnkd.in/gPQD9mhi #AIIntegration #TechInnovation #ArtificialIntelligence #OpenAI #ProjectJourney #APIDevelopment #GamingWithAI

🚀Google recently launched Gemini 1.5 - the next generation of Artificial Intelligence! 🚀 Imagine a machine that understands context like never before - that's what Google has delivered with Gemini 1.5. It's an AI model that's faster, smarter, and more powerful than anything we've seen. A true revolution in AI technology! Using a sophisticated Mixture-of-Experts (MoE) architecture, Gemini 1.5 is primed to outperform its predecessor, the Gemini 1.0 Ultra. Its speed, efficiency, and quality are truly impressive - all thanks to this structure, which combines multiple smaller expert neural networks. Don't blink, or you'll miss Gemini's 1.5's ability to process a staggering 1 million to 10 million tokens for the PRO model consistently. This means it can analyze a burst of data, everything from text to audio, to video, and even code! And there's no task too complex for Gemini 1.5, either. It's already proven its mettle, outperforming Gemini 1.0 Pro on 87% of benchmarks used for developing large language models. But it's not all about performance. Google has also prioritized ethical AI development, conducting rigorous tests to ensure representational fairness and content safety. Lastly, for my developer friends out there, Google is offering a limited preview of Gemini 1.5 Pro via AI Studio and Vertex AI. https://lnkd.in/dC88qQ87 #vertexai #googleaistudio #geminipro

Mind blowing real-time LLaMA 3 + Groq demo. Now extrapolate this 5 years out - even the technology we currently have will change the landscape of apps as we know it today quite dramatically. Not counting in daily advancements that are happening in AI. It's well known that latency changes user experience, just think of some of the research Google did in the early days showing how even a slight increase in latency leads to significantly less searches (see https://lnkd.in/df26CZam). So if we just speed up the existing capabilities the quality of future apps will increase. This realtime table manipulation is such a cool proxy for all the things we'll be able to do: e.g. imagine curating your Notion tables like this vs having to click through the GUI and google search how to do X. If you want to learn what makes Groq tick, I had their head of sillicon talk about their chips - LPUs: https://lnkd.in/dgx3kmCb And here I had Thomas Scialom, PhD LLaMA 2 author talk about LLaMA 2: https://lnkd.in/deibURjY I'll try to bring on over Thomas over the next days to share insights behind building LLaMA 3!

Latency matters and on chip inference helps tremendously. Take a look at this LLaMA3 + Groq demo now! #groq #inference #llama3 #opensource

What an amazing 48 hours in AI. While OpenAI's stunning text-to-video capability may have caught more attention, it is Google's Gemini 1.5 Pro with its 10M token context that hits closer to home for me. With near perfect retrieval from 10M tokens, many of the current design patterns to retrieve and rank the best context to add to a user's prompt (RAG) become obsolete. If you can include the codebase, requirements and design specification in the prompt and ask questions.... new use cases become possible and change everything (again). https://lnkd.in/eu58AhKY

#Topics Google's best Gemini demo was faked [ad_1] Google’s new Gemini AI model is getting a mixed reception after its big debut yesterday, but users may have less confidence in the company’s tech or integrity after finding out that the most impressive demo of Gemini was pretty much faked. A video called “Hands-on with Gemini: Interacting with multimodal AI” hit a million views over the last day, and it’s not hard to see why. The impressive demo “highlights some of our favorite interactions with Gemini,” showing how the multimodal model (that is, it understands and mixes language and visual understanding) can be flexible and responsive to a variety of inputs. To begin with, it narrates an evolving sketch of a duck from a squiggle to a completed drawing, which it says is an unrealistic color, then evinces surprise (“What the quack!”) when seeing a toy blue duck. It then responds to various voice queries about that toy, then the demo moves on to other show-off moves, like tracking a ball in a cup-switching game, recognizing shadow puppet gestures, reordering sketches of planets, and so on. It’s all very responsive, too, though the video does caution that “latency has been reduced and Gemini outputs have been shortened.” So they skip a hesitation here and an overlo...