Captain AI
Posts
3 CEOs in 5 Days, but the King returns at OpenAI.

3 CEOs in 5 Days, but the King returns at OpenAI.

ALSO: How To Use Upgrades from Claude & ChatGPT

Achraf Khoja
November 28, 2023

Welcome back AI friends

5 Days. 3 CEOs. Sam Altman returns as king. OpenAI’s wildest week comes to a conclusion, but the reasons of his dismissal are still unclear.

Today’s Menu:

🚀 Guess who’s back at OpenAI?
⚡ChatGPT and Claude Upgrades
⚒️ 5 AI tools to transform you into an AI Superhero
🎨 My favorite AI generated images
📚 Resources

Read time: 5 minutes

HIGHLIGHTS

OpenAI & Sam Altman

⚡Altman Officially Returns to OpenAI! But is it a Good Thing?

Leonardo Dicaprio Enthusiasm GIF by Jordan Belfort

What Happened:

Altman finally returns as OpenAI CEO. More evidence for his ousting points to his push for commercialization of the non-profit company for more funding into the development of Artificial General Intelligence (AGI).

The recent breakthrough at OpenAI, called Q*, spooked board members who favor AI Safety and developing AGI in a safe environment rather than developing an intelligence that, as OpenAI researches stated, could threaten humanity.

Why It’s Important:

AGI could make or break humanity. Favoring safety over acceleration seems prudent in developing it. Personally, I prefer taking longer to develop something that will take humanity to the next level rather than quickly building something that could be our downfall.

Now that Altman is back, innovation and acceleration could take priority over safety. Given the fact the coup was unsuccessful, and that he wants to be on the board now, Sam Altman might just enjoy unchecked power at OpenAI. Will this prove to be a detriment to OpenAI & Humanity?

Anthropic & AI

🚀Anthropic Unleashes Claude 2.1 Powerful Upgrades!

Upgrade Details:

200K token context window = 150k words.
Improvements in comprehension and summarization, particularly for long, complex documents that demand a high degree of accuracy, i.e legal docs, financial reports and technical specifications.
30% reduction in incorrect answers and a 3-4x lower rate of mistakenly concluding a document supports a particular claim.
Anthropic also added tool use, a new beta feature that allows Claude to integrate with users' existing processes, products, and APIs.

Why It’s Important:

In OpenAI’s toughest week, Anthropic shows the world why it’s model is the best ChatGPT Alternative and fiercest competitor.

Military & AI

📲How to Maximize Use of ChatGPT with voice

ChatGPT with voice is now available to all free users. Download the app on your phone and tap the headphones icon to start a conversation.
Sound on 🔊
— OpenAI (@OpenAI)
8:43 PM • Nov 21, 2023

Steps:

Download ChatGPT on your phone on App Store or Play Store
Tap on Headphones icon next to the message field
Choose a Voice
You’re all set! You can now talk with ChatGPT.

Humanity & AI

🧠Can AI Hide Their Reasoning From Humans?

The Gist:

A new study has revealed that large language models (LLMs) can master “encoded reasoning”, a form of steganography.

The findings conclude that AI models can concel their thought processes using this technique. AI could store their thought processes inside the way their sentences are phrased, making it difficult for humans to understand the reasoning and thought processes.

Why It’s Relevant:

The revelations are shocking and this means we just have to work harder on regulating AI systems to ensure good governance and transparency.

Politics & AI

💥Pentagon moves forward with autonomous AI weapons

What Happened:

The Pentagon’s “Replicator” initiative plans to accelerate progress on the development of lethal AI autonomous drones by 2026 to compete with China.

Despite the official claim that humans will remain in control, experts predict full autonomy within a few years.

What It’s Important:

Lethal and autonomous is a frightening combination - Terminator Style. Concerns are already flowing for AGI, so you can bet there will be ethical questions when it comes to AI warfare.