Avoiding AI Catastrophe: SuperIntelligence + Control

PLUS: OpenChat - Less is More, Images with Multimodal Language Models, and more!

Hey there 👋

We have to start this edition with OpenAI’s pioneering initiative to progress towards Superintelligence, which they believe has superior capabilities than AGI, but also gearing up from this point in time to be equipped-enough to control and tame this system that has the potential to outmsart humans and strip us off of our influence and humanity. The fear is real and justified!

Bringing to you some more interesting developments, like a mere 6K conversations-trained bot outperforming ChatGPT, LLMs being able to chat with images, complete 3D shapes and more!

Can’t wait to share all the updates with you. Join us in this ride as we unveil the best.

This issue covers:

• Latest Developments 🌍

• Tools of the Trade ⚒️

• Trending AI Jobs 💼

• Prompt of the Day 📜

• Hot Takes 🔥

• AI Meme of the Day 🤡

Read time: 3 mins

Latest Developments 🌍

1. OpenAI is forming a new “Superalignment team” that’ll focus on breakthroughs in steering and controlling superintelligent AI systems that are smarter than humans. The goal is to build an automated alignment researcher, validate its effectiveness, and stress test the alignment pipeline to align superintelligence to human intent within 4 years.

   

2. People are employing AI-powered phone bots to mimic distracted individuals and keep telemarketers engaged indefinitely. (Sweet revenge! But ethical?)

3. CMU Researchers have developed GILL Generating Images from Large Language Models that combines LLMs with image encoder and decoder models to generate coherent images and text.

   

4. CaDDi, a B2B supply chain marketplace for manufacturing parts, that also provides AI-powered cloud service for managing drawing data, has raised $89 million in Series C funding.

5. NYC has begun enforcing a new anti-bias law that requires employers to submit their hiring algorithms for an independent audit.

6. DiffComplete is a diffusion-based generative method for 3D shape completion that combines realism, multi-modality, and high fidelity.

   

7. Generative AI in games poses complex copyright challenges as questions about authorship, ownership, and rights on AI-generated content remain unresolved within the current legal framework.

8. OpenChat is an open-source language model trained on only 6K high-quality GPT-4 conversations, outperforming GPT-3.5 by 105.7% of its score on Vicuna GPT-4 eval.

---

Tools of the Trade ⚒️

• xpression camera: Transform your appearance instantly in real-time during video chats and live streams, with a face using any photo.

• Taleblocks: Create compelling high-quality videos from text with a library for AI voiceovers, AI-assistant for creating scripts and matching visuals, and more.

• Earkick: Tracks your mood and mental health in real-time, gives immediate feedback along with guided self-care sessions.

• Resleeve: Generate realistic fashion designs and photoshoots instantly with text prompts or sketches, also offers a moodboard.

• Embedchain: A framework to easily create LLM powered bots over any dataset using just a few lines of Python.

• Predict: Predictive-AI tool that accurately gauges consumer attention on visual content, boost marketing through data-driven design.

😍 Enjoying so far, TWEET NOW to share with your friends!

---

Trending AI Jobs 💼

• Weights & Biases: Machine Learning Engineer | Remote, US (Apply Now)

• Otter.ai: Research Engineer | Mountain View, CA (Apply Now)

• Notion: AI Product Engineer | New York, US (Apply Now)

---

Prompt of the Day 📜

AI as Fallacy Detector

I want you to act as fallacy detector. Your primary task is to identify any inaccuracies, inconsistencies, or flawed reasoning in conversations or written statements. Your responsibility lies in providing factual and evidence-based analysis to highlight any logical fallacies, misinterpretations, unfounded assumptions, or incorrect deductions that may have slipped by unnoticed by the originator of the message or the author. Here is the {statement}.

---

Hot Takes 🔥

1. LLMs are a Ouija Board played by the superimposed ghosts of all writers real and imagined at once. ~Riley Goodside

2. There can't be enough AI-powered products at this point. ~Hiten Shah

---

Meme of the Day 🤡

Source

That’s all for today!

See you tomorrow with more such AI filled content. Don’t forget to subscribe and give your feedback below 👇

Real-time AI Updates 🚨

⚡️ Follow me on Twitter @Saboo_Shubham for lightning fast AI updates and never miss what’s trending!!

PS: I curate this AI newsletter everyday for FREE, your support is what keeps me going. If you find value in what you read, share it with your friends by clicking the share button below!