- AI Odyssey
- Posts
- Bezos, Nvidia, OpenAI, and Microsoft join hands...
Bezos, Nvidia, OpenAI, and Microsoft join hands...
Plus: 10 latest AI developments (you won't guess #4)
Hello and welcome! Get ready to dive into the latest artificial headlines, because in this issue, we're bringing you the top news stories from the past week.
Here’s the roundup for today:
Tired of TL;DR headlines? Ex-Twitter birds build AI news reader (and fairly pays creators) 📰
Google open-sources Magika 📂
Bezos, Nvidia, OpenAI, and Microsoft join hands to ‘Figure 01’ 🤖
10 Latest AI developments 💢
4 new AI tools to experiment 🤯
AI art that will leave you in awe 🖼
AI deep deep dives 🌊
Former Twitter engineers are building Particle, an AI-powered news reader
A group of ex-Twitter engineers just launched Particle, an AI-powered news reader that promises to change the way you consume news. It claims to fairly compensate the writers as well.
The startup was founded last year by Sara Beykpour, former Senior Director of Product Management at Twitter, known for her work on Twitter Blue, Twitter Video, conversations, and the experimental app twttr, where she was from 2015 to 2021.
Here's the gist:
AI summaries: Get the lowdown on any story in quick, easy-to-understand bullet points. ⚡
Multiple perspectives: See how different sources are covering the same story with new angles, avoiding echo chambers.
Fair compensation: Particle aims to pay authors and publishers fairly for their work, unlike some other news aggregators.
Particle will include major publishers like The New York Times, CNBC, CNN, Breitbart, The Guardian, The Washington Post, and more. However, the lack of direct links to the sources in the summary makes fact-checking challenging without reviewing each article individually.
The verdict is still out. There are still some unanswered questions:
How exactly will they "fairly compensate" creators?
Can we fact-check the AI summaries easily?
While Particle is still in beta testing, it's yet to see its potential to be game-changing for news consumption. On the flip side, It’s a concern for the shrinking newsroom and AI news could decrease the ad revenue for media websites due to fewer clicks on the source.
Google open-sources Magika - AI file-type identification system
Google’s Magika helps accurately detect binary and textual file types.
Under the hood, Magika employs a custom, highly optimized deep-learning model, enabling precise file identification within milliseconds, even when running on a CPU.
Why is it needed?
For decades, computers have struggled to figure out what kind of file they're dealing with, like a confused librarian facing a mountain of unlabeled books.
This complexity is particularly pronounced in textual formats and programming languages, where similarities in constructs can confound traditional detection methods.
Current solutions like libmagic and other file-type-identification tools rely on manual heuristics and rules tailored for each format, a laborious and error-prone process.
The results: In terms of performance, Magika surpasses other tools by approximately 20% when tested on a benchmark of 1 million files covering more than 100 file types, owing to its advanced AI model and extensive training dataset.
Various file type identification tools performance for a selection of the file types
Bezos, Nvidia, OpenAI, Microsoft join hands to build a human-like robot
Tech giants like Jeff Bezos, Nvidia, and Microsoft are placing a big bet on the future of work, pouring $675 million into Figure AI, a company developing humanoid robots.
Their star creation, Figure 01, is designed to move and work just like a human. The company has said it hopes its machine, called Figure 01, will be able to perform dangerous jobs that are unsuitable for people and that its technology will help alleviate labor shortages.
Meet Figure 01
— Figure (@Figure_robot)
6:09 PM • Oct 18, 2023
Concerns arise about the potential displacement of human workers in various sectors. The question becomes: Are we preparing for a future where robots become collaborators or competitors in the workforce?
More on the funding:
Figure AI Inc., supported by OpenAI and Microsoft Corp., is raising approximately $675 million in funding with a pre-money valuation of around $2 billion. Notable investments include $100 million from Bezos through Explore Investments LLC, $95 million from Microsoft, and $50 million each from Nvidia and an Amazon.com Inc.-affiliated fund.
Latest AI hits:
Stable Diffusion 3 arrives to rival Sora and Gemini
Apple researchers Introduce Keyframer: An LLM-Powered Animation Prototyping Tool that Can Generate Animations from Static Images. (link)
Stability AI debuts Stable Diffusion 3 - most powerful yet. (link)
Paris-based AI startup Mistral AI is gradually building an alternative to OpenAI and Anthropic (link)
Tyler Perry halts $800m studio expansion after being shocked by AI (link)
Wipro and IBM team up to accelerate enterprise AI (link)
Intel lands a $15 billion deal to make chips for Microsoft (link)
Google announced that Gemini capabilities are now available across Workspace apps like Gmail, Docs, and Slides for Premium users. (link)
Chinese startup Moonshot AI raised over $1B from backers, including Alibaba, to develop LLMs handling long-form text. (link)
Disney announced its investment into five startups as part of its Accelerator program, with three of them — ElevenLabs, PrometheanAI, and AudioShake — operating in the AI sector. (link)
Perplexity announced a partnership with ElevenLabs, introducing a daily podcast narrated by the startup’s AI-generated voices. (link)
AI art to sip your coffee with…
Prompt:
**A dysfunctional neural network hallucinates a giant crystal cave of the iridescent mushroom turtles of old, stories as old as time, a galactic microbiome flourishes sputters and bubbles in gaseous pockets --s 750 --v 6.0 --ar 2:1**
Made with Midjourney
🕑It’s time to experiment the latest AI tools:
✍ Notion AI, you can automatically summarize meetings, generate topic tags, capture tasks, organize meeting notes, and more with ‘auto-fill properties’. (link)
📣 Type Prompt- Instantly generate human-like social media content with AI-driven hook prompts (link)
💼 Talently Mock Interview- AI-powered interview coach for real-time evaluation (link)
🎉 Greeting Cards GPT- Customizable themed greeting cards for any occasion (link)
💭AI deep dives: Food for thought
Still hungry for more? Here are some curated AI deep dives for you:
Meta released Aria recordings to fuel smart speech recognition: The Meta team released a multimodal dataset of two-sided conversations captured by Aria smart glasses. It contains audio, video, motion, and other sensor data. The diverse signals aim to advance speech recognition and translation research.
FlowGPT is like an "app store" for various AI models, letting users build and share their own creative AI apps, even earning tips for their work. Remember OpenAI's cool GPT Store for AI chatbots? Well, some folks want more options, so startups are stepping up!
AnyGPT is a major step towards artificial general intelligence: Researchers in Shanghai have developed AnyGPT, a groundbreaking new AI model that can understand and generate data across virtually any modality like text, speech, images, and music using a unified discrete representation. It achieves strong zero-shot performance comparable to specialized models, representing a major advance towards AGI.
Stay tuned for more. We’ll see you this Thursday...👋