AI Weekly Rundown (November 18 to November 24)
Major AI announcements from Open AI, Amazon, Microsoft, Google, Meta, and more.
Hello Engineering Leaders and AI Enthusiasts!
Welcome to another thrilling week in the AI universe! Get set for the freshest updates from the biggest players in the game. 🌟
In today’s edition:
🌐 Amazon to offer free AI training to 2 million people
🧠 Microsoft research drops Orca 2 with stronger reasoning
🚀 Runway released new features and updates
🚀 Anthropic launches Claude 2.1 with 200K context window🎥 Stability AI releases Stable Video Diffusion🔄 Sam Altman returns as OpenAI CEO
➡️ A brief timeline of how OpenAI’s dramatic week unfolded
🔍 What happened the day before Sam Altman was fired by OpenAI
🔮 Why OpenAI boardroom drama might be far from over
👊 Inflection AI’s massive 175B parameter model challenges GPT-4🗣️ ElevenLabs’s latest Speech to Speech transformation
▶️ Google Bard answering your questions about YouTube videos
Let’s go!
Amazon to offer free AI training to 2 million people
Amazon is announcing “AI Ready,” a new commitment designed to provide free AI skills training to 2 million people globally by 2025. It is launching new initiatives to achieve this goal:
8 new, free AI and generative AI courses open to anyone and aligned to in-demand jobs. It includes courses for business and nontechnical audiences as well as developer and technical audiences.
Through the AWS Generative AI Scholarship, AWS will provide Udacity scholarships, valued at more than $12 million, to more than 50,000 high school and university students from underserved and underrepresented communities globally.
New collaboration with Code.org is designed to help students learn about generative AI.
Amazon’s AI Ready initiative comes as new AWS study finds strong demand for AI talent and the potential for workers with AI skills to earn up to 47% more in salaries.
Microsoft research drops Orca 2 with stronger reasoning
A few months ago, it introduced Orca, a 13B language model that demonstrated strong reasoning abilities by imitating the step-by-step reasoning traces of more capable LLMs.
Orca 2 continues to show that improved training signals and methods can empower smaller language models to achieve enhanced reasoning abilities, which are typically found only in much larger language models. Orca 2 models match or surpass other models, including models 5-10 times larger, as assessed on complex tasks that test advanced reasoning abilities in zero-shot settings.
Results comparing Orca 2 (7B and 13B) to LLaMA-2-Chat (13B and 70B) and WizardLM (13B and 70B) on a variety of benchmarks.
Runway released new features and updates
The updates aim to provide more control, greater fidelity, and even more expressiveness when using Runway.
Gen-2 Style Presets: They allow you to generate content using curated styles without the need for complicated prompting, from glossy animations to grainy retro film stock and everything in between, Style Presets bring more styles to your stories.
Director Mode Updates: Director Mode’s advanced camera controls have been updated to allow for a more granular level of control. Now you can adjust camera moves using fractional numbers for greater precision and intention.
New Image Model Update: Improved fidelity, greater consistency and higher resolution generations are now available in Text to Image, Image to Image and Image Variation.
Add these tools to your Image to Video workflow for more storytelling control than ever before. These updates are now available to all users.
Anthropic launches Claude 2.1 with 200K context window
Claude 2.1 delivers advancements in key capabilities for enterprises– including:
Industry-leading 200K token context window, so you relay roughly 150K words or over 500 pages of information to Claude.
Significant gains in honesty, with a 2x decrease in hallucination rates compared to Claude 2.0. It has demonstrated a 30% reduction in incorrect answers and a 3-4x lower rate of mistakenly concluding a document supports a particular claim.
A new tool use feature allows the model to integrate with users' existing processes, products, and APIs. This means that Claude can now orchestrate across developer-defined functions or APIs, web search, and private knowledge bases.
Introducing system prompts, which allow users to provide custom instructions to structure responses more consistently. Anthropic is also enhancing developer experience with a new Workbench feature in the Console that makes it easier for Claude API users to test prompts.
Claude 2.1 is available over API in its Console and is powering the claude.ai chat experience for all users. Usage of the 200K context window is reserved for Claude Pro users. The pricing is updated too, to improve cost efficiency for customers across models.
Stability AI releases Stable Video Diffusion
It is Stable Diffusion’s first foundation model for generative video based on the image model Stable Diffusion. It is adaptable to various video applications and is released as two image-to-video models. At the time of release in their foundational form, through external evaluation, these models surpassed the leading closed models in user preference studies.
Now available in research preview, It is not yet ready for real-world or commercial applications at this stage.
Sam Altman returns as OpenAI CEO
OpenAI has reached a tentative deal to allow for Sam Altman to return as the company’s CEO and form a new board of directors.
Co-founder Greg Brockman will also be returning to the company, days after stepping down as president in response to Altman’s firing.
The initial board has been put in place to “vet and appoint” a full board with up to nine members. Altman has reportedly sought a place on the new board, and so has Microsoft– the biggest investor in OpenAI. In addition, the company will investigate Altman’s controversial firing and the subsequent drama.
Enjoying the weekly updates?
Refer your pals to subscribe to our newsletter and get exclusive access to 400+ game-changing AI tools.
When you use the referral link above or the “Share” button on any post, you'll get the credit for any new subscribers. All you need to do is send the link via text or email or share it on social media with friends.
A brief timeline of how OpenAI’s dramatic week unfolded
November 16
The OpenAI board receives a letter from researchers warning of a powerful AI discovery that they said could threaten humanity. Reuters reported today that the initial ousting of Altman may have been sparked by this letter.
November 17
OpenAI board fires CEO, co-founder Sam Altman
President Greg Brockman quits after being removed from OpenAI’s board of directors
CTO Mira Murati is named interim CEO
November 18
Sam Altman is reported to be planning a new AI venture and Brockman is expected to join the effort
Some OpenAI employees contemplate quitting if Altman is not restored as CEO, and some expressed support to joining his new venture
Major investors pressure OpenAI board to reverse the decision
If Altman returns, the board will be replaced
Microsoft CEO Satya Nadella urges OpenAI board to reconsider bringing Altman back as CEO
November 19
OpenAI rivals rush to poach OpenAI employees
Sam Altman discusses a possible return to the company
After a weekend of negotiations, Altman will not return
Emmett Shear, co-founder of Twitch, will take over as interim CEO
Dozens of staffers quit OpenAI
November 20
OpenAI staff revolt raises pressure on board to back down
Altman, Brockman, and other OpenAI employees would join Microsoft to lead “a new advanced AI research team”, says Microsoft CEO Satya Nadella
Nearly all of OpenAI's 700+ staff threatens to join Altman at Microsoft if he wasn't reinstated
More than 100 OpenAI customers consider defecting to rivals like Anthropic, Google, Microsoft
OpenAI board approaches Anthropic about a merger but is turned down
November 21
Sam Altman is reinstated as OpenAI CEO, Brockman also returns and there will be an internal investigation
A new initial board is formed, chaired by Bret Taylor, a former co-CEO of Salesforce, with former Treasury Secretary Larry Summers and Adam D'Angelo as other members
What happened the day before the OpenAI board fired Altman?
Before OpenAI CEO Sam Altman’s four-day exile, several staff researchers wrote a letter to the board of directors warning of a powerful AI discovery that they said could threaten humanity.
The letter was a key factor on a long list of grievances against Altman. The list also had concerns over commercializing advances before understanding the consequences.
While OpenAI declined to comment about this, it had acknowledged in an internal message to staffers a project called Q*. Some at OpenAI believe Q* (pronounced Q-Star) could be a breakthrough in the startup's search for artificial general intelligence (AGI).
Check out the Reuters report for more details.
Why OpenAI boardroom drama might be far from over
Despite the above report, the big question still remains unanswered: what happened between the board and Altman that led to his firing?
And Sam Altman faces the hard part– mending all the fractures that led to his ouster and followed it.
Among the most immediate challenges will be determining the role of Ilya Sutskever, the company’s visionary chief scientist, and his allies on the company’s AI safety team who initially supported Altman’s ouster. At the same time, Altman needs to act quickly to undo any damage to OpenAI’s standing with its customers and employees. Plus, there are reported tensions between Sam Altman and Adam D’Angelo. It is also not certain what the final shape of the new board will look like. And so on.
We’ll be keeping an eye on these unknowns for you.
Inflection AI’s massive 175B parameter model challenges GPT-4
Inflection AI has released the Massive 175B Parameter Model- Inflection-2. It is the latest language model developed by Inflection, aiming to create a personal AI for everyone. It has been trained on 5K NVIDIA H100 GPUs and demonstrates improved factual knowledge, stylistic control, and reasoning abilities compared to its predecessor, Inflection-1.
Despite being larger, Inflection-2 is more cost-effective and faster in serving. The model outperforms Google's PaLM 2 Large model on various AI benchmarks. Inflection takes safety, security, and trustworthiness seriously and supports global alignment and governance mechanisms for AI technology. Inflection-2 will undergo alignment steps before being released on Pi, and it performs well compared to other powerful external models.
ElevenLabs’s latest Speech to Speech transformation
The company has added STS to Speech Synthesis, allowing users to convert one voice to sound like another and control emotions, tone, and pronunciation. It can extract more emotions from a voice or be used as a reference for speech delivery.
Changes are also being made to premade voices, with new ones added and information on voice availability provided. Other updates include the addition of normalization, a pronunciation dictionary, and more customization options to Projects. The Turbo model and uLaw 8khz format have been introduced, and ACX submission guidelines and metadata can now be applied to Projects.
Watch this video created by one of their community members:
You can try it here.
Google Bard answering your questions about YouTube videos
Google's Bard AI chatbot can now answer specific questions about YouTube videos, expanding its capabilities beyond just finding videos. Users can now ask Bard questions about the content of a video, such as the number of eggs in a recipe or the location of a place shown in a travel video.
This update comes after YouTube recently introduced new generative AI features, including an AI conversational tool that answers questions about video content and a comments summarizer tool that organizes discussion topics in comment sections.
That's all for now!
Subscribe to The AI Edge and gain exclusive access to content enjoyed by professionals from Moody’s, Vonage, Voya, WEHI, Cox, INSEAD, and other esteemed organizations.
Thanks for reading, and see you on Monday. 😊