Crossover ’s Post

"I've only had it for like a week or something," he [Sam Altman] said, but one way of using it is "putting my phone on the table while I'm really in the zone of working, and then, without having to change windows or change what I'm doing, using it as like another channel," he said. He elaborated, "So I'm working on something, I would normally stop what I'm doing, switch to another tab, Google something, click around or whatever, but while I'm still doing it, to just ask and get an instant response without changing from what I was looking at on my computer, that's been a surprisingly cool thing." This is the very definition of "ubiquitous computing" and something I do all the time using my Google Home speaker. I will be doing that for much more than simple tasks such as checking my spelling as soon as I can talk to Gemini from it. Early on in my career I read a scholarly article about ubiquitous computing that made the argument that the power of computing needs to become so prevalent that it essentially disappears. It made the point that eye glasses are a technology that extends the power of your sight but when you put on a pair, you don't look at them, you look through them. Computers need to become ambient and effortless. We're almost there. #AI #ArtificialIntelligence

A significant portion of Google’s success, the digital behemoth known for innovation and quality, may be attributed to the thoughtful integration of machine learning algorithms throughout its wide range of goods and services. Google uses machine learning to improve user experiences and propel corporate growth in a variety of areas, including search algorithms and personalised suggestions. The top ten machine learning algorithms that Google uses to achieve its amazing feats are examined in this article. https://lnkd.in/g55am9Yz #bert #DeepMind #AlphaGo #google #MachineLearning #PageRankAlgorithm

Google's Gemini is here !!!!! 🎊 Gemini is Google's most capable model with sophisticated multimodal reasoning capabilities. Designed for flexibility, Gemini is comes in three different sizes — Ultra, Pro and Nano — so it can run on everything from data centers to mobile devices. From natural image, audio and video understanding to mathematical reasoning, Gemini Ultra’s performance exceeds current state-of-the-art results on 30 of the 32 widely-used academic benchmarks used in large language model (LLM) research and development. With a score of 90.0%, Gemini Ultra is the first model to outperform human experts on MMLU (massive multitask language understanding). Starting on December 13, developers and enterprise customers can access Gemini Pro via the Gemini API in Google AI Studio or Google Cloud Vertex AI. Gemini Nano will now run on Pixel 8 Pro. A specifically fine-tuned version of Gemini Pro will be available in Bard from today onwards. Gemini Ultra will come to Bard early next year in a new experience called Bard Advanced.

The major players in the AI world are staying busy. 😅 This month, OpenAI, Google, and Microsoft all unveiled exciting new models and capabilities. Discover all the details in this jam-packed episode of AI This Month. 👏 🤖 Microsoft unveiled Copilot+ model and PCs 🤖 GitHub announced GitHub Copilot Extensions 🤖 Azure AI Studio generally available 🤖 OpenAI announced the release of GPT-4o 🤖 New Google Models: Gemini 1.5 Flash, Veo, Imagen 3, and AlphaFold 🤖 Microsoft Phi-3 family expands, works on MAI-1 🤖 IBM's Granite Code Models are now open source 🤖 Anthropic’s LLM research 🤖 Google introduces SynthID The next wave of #ai models is here. 🌊

Google is trying to keep up with all the latest *BIG* changes that are happening and here are some announcements from their I/O day which are worth sharing in the race to lead Generative AI space ▶ 𝐍𝐞𝐰 𝐓𝐏𝐔 (𝐓𝐫𝐢𝐥𝐥𝐢𝐮𝐦) -> If this project will become successful this might become the biggest and most impactful project from Google in the recent years. As we see who important the compute is and big players like Google would love to have as less dependency on the like of NVIDIA and other related to this. ▶ 𝐕𝐄𝐎 -> A new text to video model which is Google's answer to the SORA model from OpenAI ▶ 𝐈𝐦𝐚𝐠𝐞𝐧 3 -> Pretty straightforward - a better version of the previous Imagen versions ▶ 𝐏𝐫𝐨𝐣𝐞𝐜𝐭 𝐀𝐬𝐭𝐫𝐚 -> This is introduced as your assistant with Multimodal capabilities. Long way to go however, this was for sure my favourite one. It is become pretty clear that the future in this fields is in (1) Multi-modalities and (2) Agents. Eventually, we would all like to use these tools more like personal assistants to increase our efficiency and this is what Google is hinting here. ▶ 2𝐌 𝐜𝐨𝐧𝐭𝐞𝐱𝐭 𝐥𝐞𝐧𝐠𝐭𝐡 𝐟𝐨𝐫 𝐆𝐞𝐦𝐢𝐧𝐢 -> Well, another RAG killer? We will see. _____________________________________ ✅ Follow me Naqqash Abbassi for regular updates and insights on Generative AI and my journey as a CTO in AI product development

A little late in the evening, but wanted to share my #AITuesday post: 1. Princeton open sourced SWE-agent SOTA with 12% fix rate for Github issues 2. Cloudflare announced running fine-tuned LoRA models on Workers AI 3. Apple researched develop AI that can see and understand screen content 4. Apple says their latest ReALM model is better than GPT4 5. Lumos is learning agents with unified data, modular design, leveraging open source LLMs

Check below for a snapshot into what the most significant players in the AI world are up to!

The increasing scale and complexity of modern ML and AI models (e.g., Transformers and diffusers) pose significant challenges to existing visualization tools. Large-scale Transformer-based models often overwhelm conventional graph visualizers, resulting in rendering failures or unmanageable visual complexity. In addition, the lack of a clear visual hierarchy complicates one’s understanding of node relationships. To address these limitations, we introduce Model Explorer, a novel graph visualization solution that can handle large models smoothly and visualize hierarchical information, like function names and scopes. Model Explorer supports multiple graph formats, including those used by JAX, PyTorch, TensorFlow and TensorFlow Lite. Developed originally as a utility for Google researchers and engineers, Model Explorer is now publicly available as part of our Google AI Edge family of products. 🌐 Read the announcement: https://lnkd.in/eb7ZTdBA 🤖 Visit the Model Explorer site for installation instructions, and to provide feedback: https://lnkd.in/ejHEZrra #GoogleIO

Our next newsletter is on “Why We Want (and Need) Virtual Assistants.” With the release of OpenAI’s GPT-4o and the announcement of Google’s Project Astra, it is clear that tech thinks we need them. But do we want and need them? The answer we think is a hearty, “Yes!!” Read on to find out why.   The link to subscribe and read our newsletter is in the reply below.

Google - either showing you what THEY will have and WE won't or it's all coming this way from them WITH THEM AT THE TOP -- multi-modal Ai -- text, images, audio and more - "𝐀𝐋𝐋 𝐚𝐭 𝐭𝐡𝐞 𝐬𝐚𝐦𝐞 𝐭𝐢𝐦𝐞" + Reasoning + Advanced coding https://lnkd.in/gGVj6BBm 𝐋𝐎𝐎𝐊 𝐚𝐭 𝐭𝐡𝐞 𝐩𝐢𝐜𝐭𝐮𝐫𝐞 𝐢𝐧 𝐭𝐡𝐞 𝐚𝐫𝐭𝐢𝐜𝐥𝐞 𝐨𝐟 "A row of Cloud TPU v5p AI accelerator supercomputers in a Google data center" all the while it can also make 🥓 🥞 ☕ and slices 🍞 while you wait. 😎 #ai #aiadvancements #businessautomation #MobiTextVerts