Daily Digest: We're so back

PLUS: Anthropic strikes, new video model

November 22, 2023

Sign up | Advertise | Ben’s Bites News
Daily Digest #290

Hello folks, here’s what we have today;

PICKS

  1. The drama at OpenAI seems to have come to a halt for now. Sam is returning as CEO. The OAI team got back to shipping even before the news broke. ChatGPT with voice is now available to all free users. 🍿Details and what’s next (also below)

  2. Anthropic has released Claude 2.1, the latest version of its AI assistant, with several major upgrades including a 200,000 token context window, reduced rates of hallucination, and new tool use capabilities. So with the new chad in town, will you break up with OpenAI?🍿Our Summary (also below)

  3. Stable Video Diffusion - Stability AI has released its first foundation model for generative video based on the image model Stable Diffusion. Stable Video Diffusion can generate 14 or 25 video frames at customizable frame rates.

from our sponsor

WANT TO GET REAL RESULTS WITH AI... BUT NOT SURE WHERE TO START?

At the end of the day, all you care about is generating leads, maximizing sales and improving support.

Learn step-by-step how to scale your small businesses with A.I., even if you have zero experience or technical skills.

It's time to finally put AI to work for you.

TOP TOOLS

  • Dear AI - Generate intimate, thoughtful, and beautiful letters.

  • RAGs - ChatGPT-like chatbots on your data with Llamaindex.

  • Flawless - Instant UX audit for your landing page.

  • Tuna - Rapidly generate synthetic fine-tuning datasets.

  • Hear the web - Instantly convert your newsletter into an engaging podcast.

  • Audio Writer - Your thoughts, our words. It's magic.

  • VoiceGPTs - Shareable voice bots that you can use in seconds.

  • OpenAI load balancer - Distribute API requests across multiple endpoints.

  • Lifeboat by Replicate - Switch your language model provider to Llama 2 70B with 3 lines of code. (free for this week)

  • Trace - Design your app’s interface in minutes.

  • LM Studio - Discover, download, and run local LLMs.

  • Enter a URL to clone a live website with HTML and Tailwind.

WHO’S HIRING IN AI

  • .txt - Make text generation fast and reliable.

NEWS
QUICK BITES

The drama at OpenAI seems to have come to a halt for now. Sam is returning as CEO. The OAI team got back to shipping even before the news broke. ChatGPT with voice is now available to all free users.

Nice. I want details…

The OpenAI X account posted that they have reached an agreement in principle. Sam will return to OpenAI as CEO with a new initial board of Bret Taylor (Chair), Larry Summers, and Adam D'Angelo.

Taylor is ex-CEO of Salesforce and one of the most senior/sought-after folks on tech company boards. Summers is the former US Secretary of the Treasury.

Bloomberg reported that the final discussions on who’ll be on the board were sticky. OpenAI's initial board will also vet and appoint a new formal board of up to 9 people. Microsoft and Sam Altman will likely have a seat on that formal board. (The Verge)

And the rest of the people?

Emmett is now the Interim ex-CEO of Open AI as per his X bio. He’s glad to have played a part in the solution.

Though Helen Toner (one of the coup* members) posted “And now, we all get some sleep” retweeting the OpenAI announcement, the NYT reported that a paper co-written by her in October was a major part of the disagreements.

There’s still one big question: why did all this happen?

The new board will investigate this internally, to which Sam has agreed. A few hours before the “All is well” news came, a bunch of former employees of OpenAI shared a letter to the board to expand the scope of the investigation with their allegation of Sam’s (and the exec team’s) actions during the 2019 transition from non-profit to for-profit.

So, if you’re drama-hungry, there’s more to happen. And if you’re a late bloomer and don’t know what happened, check out our two pieces covering: 🍿the weekend AI stopped and 🍿the updates leading up to the return of OpenAI.

QUICK BITES

Anthropic has released Claude 2.1, the latest version of its AI assistant, with several major upgrades including a 200,000 token context window, reduced rates of hallucination, and new tool use capabilities. So with the new chad in town, will you break up with OpenAI?

What is going on here?

Anthropic has launched Claude 2.1, with a set of improvements over Claude 2.

What does this mean?

200,000 tokens are equivalent to 500 pages of text, so you can fit all the messages you send to your ex. It’s also improving in using that big context window—2x fewer instances of hallucinating information. Anthropic calls this gains in honesty. Feels like your ex saying, “Yes, baby, let’s get back together. I’ve made significant gains in honesty.”

Claude 2.1 is also more accurate and reliable in its outputs—a 30% drop in incorrect answers and 3-4x lower rates of unsupported claims. In simple words, Claude will lie less and try not to gaslight you.

With API access, you can now also access extra tools via API, but the feature is in beta. Some more features in the developer console to top it all off: System prompts and a playground for testing prompts.

Why should I care?

For enterprise users of Claude's API, version 2.1 brings the assistant to new levels of functionality with full documents. New tool integrations allow Claude to directly query databases, make API calls and more based on natural language requests.

Claude 2.1 is live on the free and pro plans but the increased context length is reserved for pro users.

Ben’s Bites Insights

We have 2 databases that are updated daily which you can access by sharing Ben’s Bites using the link below;

  • All 10k+ links we’ve covered, easily filterable (1 referral)

  • 6k+ AI company funding rounds from Jan 2022, including investors, amounts, stage etc (3 referrals)

Reply

or to participate.