🤖 Elon Musk Begins OpenAI Battle

PLUS: Snapchat Wants In On The AI Fun

small-banner

What's up? You're reading Inclined AI. Check-in on your local morning show. They’re not doing okay, you guys.

You’ll find these stories in today’s edition:

  • Elon is in talks to establish an OpenAI competitor

  • Snapchat unveils its chat AI tool

  • Tencent setups a new team for ChatGPT-like creations

  • ChatLLaMA is already showcasing LLaMA’s capabilities

ELON MUSK TIPTOES AROUND STARTING A NEW RESEARCH LAB

Like every other newsletter to ever exist, we find ourselves covering Elon Musk for some groundbreaking, industry-quaking news.

Jon Victor and Jessica Lessin, journalists for The Information, are back with another enormous AI scope. They wrote yesterday that Elon Musk is in talks to start an OpenAI competitor.

Elon was part of the founding team of OpenAI in 2015. However, he left in 2018 when the company started its “for-profit” arm.

The report indicates that human bias in AI and constant errors from ChatGPT motivated Elon to step off the sideline and get back in the race.

Elon’s Hinted At This For Weeks

Anytime he mentions OpenAI, Microsoft, or GPT in a post, it’s been to critique the companies. Elon also worries about the development of AI. He thinks we’re playing toward a bad ending in the video game called life.

Don’t get it twisted though; Elon is not doing this out of fear or greed.

There’s a huge political element to this decision, but you come here for the facts, not the speculation.

Igor Babuschkin, one of Elon’s top targets and former researcher for Alphabet’s DeepMind AI lab, went on record about the nascent project saying, “The goal is to improve the reasoning abilities and the factualness of these language models.”

To translate, they’re not setting out to make a chatbot with fewer content safeguards. Instead, they will tackle bias, but precautions of some sort will be necessary.

Elon’s hinted at one problem he’s bound to tackle if he uses this lab to build an LLM: Reinforcement Learning from Human Feedback (RLHF). A method popularized by OpenAI right around the time Elon left. 

RLHF uses human feedback to adjust models for the desired responses. Critics, like Musk, claim this concept leads to human error and bias in ChatGPT.

Regardless, we’re in the early stages here. Another lab with significant financial backing that looks for new ways to train models is not a bad thing.

My only question is, does Elon create this as a branch of one of his current companies like Twitter, or does he add an entirely separate entity to his growing plate?

Let’s see.

Prompt: a robot inside a mobile phone Style: Pixar

ALLOW ME TO INTRODUCE YOU TO “MY AI”

Snapchat released a new chatbot for premium subscribers built off OpenAI’s GPT tech.

The name, My AI, does not inspire confidence. Funny enough, that’s also the theme of this entire take. None of this brings the hype.

Snapchat trained the model on their guideline and safety measures, hoping it’s enough to deter bad behavior.

The Company’s Goal is to Keep My AI Light & Fun

There’s no search engine component here. Instead, users can ask for fun trivia and pass the time. That idea sounds like a company betting on its distribution to secure a rushed product drop’s success.

Why did Snap even bother pushing this out?

The answer is simple: they had to respond. All the large tech companies are starting to offer these tools, and Snapchat’s users skew on the younger side.

Nothing shouts, “I’m old and not hip,” more than failing to keep up with AI.

I’m sure Snapchat also wants some sweet, sweet conversions to their premium subscription, too, but that’s not the long-term strategy.

They’ve managed to mark themselves as the first big player to offer a quick, mobile-friendly chat AI built inside their app. ChatGPT isn’t to that point yet, and Meta told users to do it themselves.

When push comes to shove, that might make a big difference in how much data they gather compared to other companies.

Quick Nuggets

🗽 NYC will regulate AI hiring, but critics say the move is lackluster

💼 Robin AI raises $10.5m Series A to support its generative AI legal tooling

🚘 Car ads and automotive retail enjoy a nice AI facelift 

🪰 Fly Tipping is a growing gross problem in the UK. AI in CCTV is attempting to help

🌴 Jordan’s date palms are saved. An AI app helps detect a nearly invisible pest

🇨🇳 Tencent adds its name to the list of Chinese companies building ChatGPT-like products

👀 Anthropic is starting to supply startups with its text-generating AI models — BFD 🌟

❓ A new chatbot was made to be stupid on purpose

🦙 ChatLLaMA is the first open-source creation built from LLaMA’s Reinforcement Learning from Human Feedback

📺 John Oliver gave his take on AI and its direction moving forward

📜 Mesopotamians can teach us about new tech and AI’s surge

📻 RadioGPT is the first fully AI-run radio station

🧬 AI transforms the way we understand genomics while unlocking new possibilities in the field

🔥 Tinder messages written by ChatGPT can come off as creepy

🤖 Mark Zuckerberg announced plans for AI products to be added to WhatsApp and Instagram

🔥 Fresh Products

  • Hypertype - write smarter emails using company intelligence (link)

  • Cosonify - find and make new songs (link)

  • Fama.one - AI-powered crypto investment vaults (link)

  • Galatical AI - AI results that fit your needs (link)

  • Mememorph - turn yourself into your favorite meme (link)

  • Magic Stickers - AI helps create personal iMessage stickers (link)

  • SupaRes - blazingly fast AI image enhancement (link)

  • Shownotes - automagically summarize podcast episodes (link)

  • mymind - Remember everything. Organize nothing. (link)

  • Toucan - fully-customizable AI writing assistant (link)

  • Okaaaay - create unmatched ad copy (link)

  • Lemonaid Music - get a fresh perspective on music creation (link)

Good Content, Total BS

Looks cool, right? Too bad the timecodes are off, and some of this never happens, either. Part of me wonders if the act of confidently making up knowledge of movies makes ChatGPT the ultimate cinephile 🤔

That’s it for today. I hope you enjoyed the latest edition of inclined.ai - Davis.

Want more?

Upgrade today for less than the price of a latte a month and get benefits like:

  • A weekly Sunday deep dive into AI trends

  • Access to Q&A privileges to get your biggest AI questions answered

  • Exclusive discounts & invites to beta AI tools

Headline for Next Sunday: Everything We Know About GPT-4 🙂