🤖 Guess who's back

PLUS: Bard’s Youtube integration allows you to ask it questions

Haroon ChouderyNov 24, 2023 · 10 min read

New here? Welcome to NAB, and make sure to subscribe to stay informed on the latest developments in AI!

Happy Friday, fellow humans 👋

Hope you had a great time celebrating Thanksgiving surrounded by friends and family 🦃

I’m thankful for you, and I appreciate you tuning in to read NAB more than you'll ever know :)

Let’s dive into some AI news…

Have anything you’d like to share with over 40k AI enthusiasts?

Sponsor Not A Bot

🗞️ Your Weekly Round-Up

1️⃣ Guess who’s back

Alright, where do we begin…

Since our last edition on the OpenAI saga, some pretty interesting things have happened:

On Sunday night / Monday morning, Sam Altman and Greg Brockman were hired by Satya Nadella to run Microsoft’s AI research unit.
Emmett Shear, former CEO of Twitch, was hired to replace Sam as OpenAI’s new CEO.
Over 700 of the 770 employees at OpenAI signed a petition that essentially read “the board walks, or we do.”
Shear started an inquiry into why Altman was fired and couldn’t get definitive answers from the board. He then decided to turn heel (or …face?) and negotiate with the board to bring Sam back.
72 hours into his tenure, he brought Sam Altman back to the helm at OpenAI with a new board in place. We knew this was going to happen. But 5 days? That’s got to be a record.

During this back and forth, some intriguing things were brought up that potentially point to reasons why Sam was terminated:

A few weeks before his dismissal, Sam disagreed with board member Helen Toner over her critical research paper, which praised Anthropic's approach to AI safety while questioning OpenAI's efforts. This discord within the board contributed to Sam's firing.
Another pivotal factor could have been the revelation and advancement of OpenAI's Q* project, hailed as a potential leap in achieving Artificial General Intelligence (AGI). It looks like the board disapproved of this initiative, leading to Sam's dismissal.

I’m sure this is not the end of the saga, but as usual, I will keep you updated :)

It’s also worth mentioning that amidst the leadership shake-up, Greg Brockman announced ChatGPT Voice, a feature that lets users converse with ChatGPT in a human-like, conversational experience.

Powered by OpenAI's Whisper technology, the feature comes in five distinct voices—Juniper, Sky, Breeze, Ember, and Cove— with Brockman claiming it 'totally changes the ChatGPT experience.’

Kudos to the OpenAI team for releasing this during the craziest week of their lives!

Read more: Gizmodo

2️⃣ Inflection AI 2 - bigger and better

Inflection AI just announced Inflection-2, a 175-billion parameter model that drastically advances their quest for a personal AI for all.

Here are the deets:

Inflection-2 has enhanced factual knowledge, superior stylistic control, and remarkable reasoning abilities.
Trained on 5,000 NVIDIA H100 GPUs, it is the second most powerful LLM globally, competing head-to-head with Google's PaLM 2 Large.
The transition to H100 GPUs and optimized inference implementation allowed for increased speeds and reduced costs.
Inflection-2 excels in tasks from MMLU and HellaSwag (common sense reasoning benchmarks) to coding benchmarks.
Despite its massive size, it outperforms leading external models like Claude-2, LLaMA-2, and Grok-1.

Read more: Inflection Blog

3️⃣ Stability AI gets into the video-generating game

Stability AI announced Stable Video Diffusion, a cutting-edge AI model that transforms stills into high-quality videos with applications spanning education, creativity, design, and more.

Stable Video Diffusion comes in two versions—SVD and SVD-XT—delivering seriously impressive four-second clips from stills.

While it’s a relatively new model, there are still some limitations. The model:

Cannot generate videos without motion or slow camera pans,
Cannot be controlled by text, and
Cannot consistently generate faces and people "properly."

There are also concerns regarding the model’s misuse, especially given the lack of built-in content filters.

It will be interesting to see the guardrails that Stability AI places to ensure that the tech is used correctly and not misappropriated for wrong or illegal use.

Read more: Yahoo Finance

4️⃣ Bard’s Youtube integration allows you to ask it questions

Google improved Bard by leveling up its YouTube integration, allowing users to analyze videos to provide key information without ever needing to hit play.

Here’s why it’s a good feature:

Users can input video links into Bard and ask the chatbot questions to clarify parts of the video or get summaries.
The use cases extend to the classroom, where students can interact with pre-recorded lessons, allowing for active learning.
Caveat: users would have to watch the video first.

It also raises concerns:

Content creators may lose out on engagement because Bard skips the step of watching, potentially impacting revenue streams tied to ads and recommended videos.
It’s not perfect. In a test with a cocktail recipe video from America's Test Kitchen, Bard accurately extracted key details but missed the video's nuances.

Currently, this feature is part of an opt-in Google Labs experience, posing a minimal threat to YouTube creators. However, the prospect of its integration into the platform does bring up interesting conversations about the balance between user convenience and creator compensation.

Read more: The Verge

5️⃣ Eleven Labs launch Speech to Speech

Let’s keep this one light and breezy.

Eleven Labs launched Speech to Speech (STS) in Speech Synthesis, a feature that allows users to transform one voice into another and control emotions, tone, and pronunciation beyond typical Text-to-Speech (TTS) capabilities.

Current use cases include:

Games,
Videos,
Podcasts, and more.

Read more: Eleven Labs Blog

🎙️ New Episode ft. Paul Yacoubian, CEO at Copy.ai

Hey team! ICYMI, our latest BWAI episode featuring Paul Yacoubian, Founder and CEO at Copy.ai, is live now!

Catch this episode on YouTube, Spotify, or Apple Podcasts today!

🤑 AI Fundraising News

Malou raises $10M in funding to help restaurants improve their online reputation, visibility, and customer engagement by integrating their Google page, social media profiles, and listing and delivery platforms into a centralized hub.
AI21 raises $208M in Series C funding to create a developer platform for businesses to build AI-driven applications and services and a consumer product called Wordtune, an AI-based reading and writing assistant.
SeeChange raises $10M in funding to use AI vision to help retail outlets from warehouse to checkout.
Climate Alpha raises $5M in Seed funding to use AI-driven analytics to deliver a comprehensive solution for navigating accelerating climate volatility in the real estate, asset management, and insurance industries.
DoorFeed raises €12 Million in Seed funding to institutionalize residential real estate investments by managing customized portfolios of apartments, houses, and buildings for investors using a mix of AI, ML, and its network of local partners.

🛠️ AI Tools to Check Out

(Sponsored) Rewind: A personalized AI powered by everything you’ve seen, said, or heard. Rewind can record, transcribe, and summarize all your meetings without an awkward bot joining the meeting. It's like having notes of your entire day without actively taking them! Used mostly by founders, CEOs, managers, executives, and directors to be wildly more productive throughout their day. Try Rewind for free Today!
Fork.ai: Find sales leads with contact info by searching tech stacks. Search apps based on technology usage and generate a tailored list of sales leads with detailed company & contact info. Check them out!