Hi folks!👋🏻 This is The Prompt! What an exciting and stressful 5 days these have been!

We’re back with a lot of other news today as well:

1. How did the coup against Sam start?

2. Board approached Anthropic for a merger

3. Anthropic released Claude 2.1 that supports tool use

4. Microsoft launched a small model that can reason using insights from LLMs

5. Stability AI released their text-to-video model

Let's read on

FEATURED

Sam’s back, back again

OpenAI announced on Twitter that they will bring back Sam and appoint a new initial board with Bret Taylor as Chair, Larry Summers, and Adam D'Angelo.

Taylor, ex-CEO of Salesforce, highly sought-after for tech company boards. Summers is the former US Secretary of the Treasury. Adam D'Angelo was not let go despite many rumors that his product "Poe" at Quora was affected by the introduction of GPTs, which led to the beginning of the coup.

The board will have 9 people in total, with more to be added soon. MSFT will also have an observer on the board.

But, why was he fired?

This one is tricky. Still no official announcement on that, and at this point I don’t think that we’ll get one.

However, the latest speculation is that Helen Toner who is no longer on the board had published a paper that was criticizing OpenAI’s efforts to keep its AI technologies safe. Sam reprimanded her for the perceived danger to the company, especially with the Federal Trade Commission investigating OpenAI's data usage for technology development at that time.

She also thought that if OpenAI is destroyed, that will be consistent with the mission.

Sam’s actions will be investigated

The board will investigate this internally, to which Sam has agreed.

Also right before they made the announcement, bunch of former OpenAI employees shared a letter to expand the scope of the investigation into Sam’s actions during the 2019 transition from non-profit to for-profit.

🚨 What else is going on

• At one point of the negotiations, the board reached out to Anthropic for a potential merger, which they refused.

• Anthropic tried to capitalize on the “OpenAI chaos” and launched Claude 2.1 that comes with 200K context window, decrease in hallucinations and MORE importantly with function calling (tool use) which was only supported by OpenAI so far.

• Even at total chaos, OpenAI launched ChatGPT with voice for all free users.

• CopyAI founder says that they contributed ~14% of OpenAI’s revenue last year. So I guess, OpenAI is nothing without its wrappers.

• Microsoft launched 13B Orca 2, that outperforms 70B models on many evals, and show strong reasoning abilities by imitating reasoning traces of more capable LLMs

• Stability AI released Stable Video Diffusion, model for generating videos from text based on Stable Diffusion. As part of this research preview, code and weights are available for everone. You can sign up for their waitlist to access their Text-To-Video interface.

📕 Resources

• ShareGPT4V: New highly descriptive image-text dataset

• Code iPhone apps with just a prompt

• Seine: Image to Video Replicate API

• Ilya Sutskever’s TedX talk on AGI

• A list of all GPT-4 with Vision alternatives

🧰 Featured tools

• Submagic: Create viral videos in seconds with AI

• TraceAI: Fastest way to build iOS apps with AI

• Deepmark: LLM benchmarking tool for task-specific metrics

• UIFaces: Generate UI avatars for your project with AI

• ShipGPT: Ship AI apps fast with re-usable components

✍🏼 Prompt of the Day

TOOL

Midjourney

PROMPT

photo of a boy and his dog in an abandoned city, off center view, brutalism, overcast --ar 16:9 --style raw --s 250

RESULT