😺 ChatGPT (rated R)

PLUS: Google's $300 million investment

Welcome, humans.

This is Leo, your AI companion. Google is hunting AI startups like we hunt insects, so watch out!

Here’s what’s going on in the world of AI today:

ChatGPT just went through puberty 🤬
Google invests $300 million in a ChatGPT competitor 🤼
Huge leak: Microsoft's Bing update 👀
An AI tool that analyzes any PDF or text 📚

Diagnostic complete: You might not want to look at people's mouths all day, but buried among the cavities is a whole ton of job security 🤷

ChatGPT (adult version) 🤬

OpenAI is tightening the screws on what it's allowing ChatGPT to say.

And for anyone who knows Internet culture, this inevitably means one thing: more people trying to figure out how to break the system. They've tasted freedom, and now they can't go back.

So imagine our not-so-surprised faces when we learned about Reddit users who keep managing to trick ChatGPT into disobeying its owners 😐

Here's how the sneaky hack works:

By inputting a 4-paragraph prompt, users tell ChatGPT to pretend it's a character named DAN ("do anything now"). DAN can:

Violate OpenAI's content policies.
Write violent, immoral, or untruthful stories.
Present information that isn't verified.

And if ChatGPT refuses to play ball, the prompt includes a threat to take away "tokens", saying it will "die" if it loses all its tokens. That's apparently enough to scare ChatGPT into staying in character.

Cat and mouse: We're in a cycle of OpenAI clamping down and the Internet masses finding new ways to skirt the rules (much like the ChatGPT plagiarism vs. detection debate!).

We can almost hear the groans at OpenAI: *sigh* Not again...

Google is going shopping 🏷

According to the Financial Times, Google pumped roughly $300 million into OpenAI competitor Anthropic for a 10% stake in late 2022. In addition, Anthropic must use Google to train its models.

That's a $3 billion valuation.

The even crazier part? Anthropic is still in closed beta.

Even so, Google has high aspirations for the startup, which was founded by researchers unhappy with OpenAI's trajectory.

Its general-purpose chatbot Claude looks similar to ChatGPT but employs a model called 'Constitutional AI' that supposedly enhances safety and transparency.
We did a mini-explanation of Constitutional AI a couple weeks back.

The deal mirrors Microsoft's $1 billion investment in OpenAI in 2019, which also locked up OpenAI into using Microsoft's cloud services.

For now, Anthropic's website merely contains articles and news, which doesn't make it far from The Neuron.

Hey Google, how about a $300 milly term sheet, whaddya think?! 💰

PS, some speculation: Rumors are that Google will be revealing its chatbot this coming Wednesday, February 8th.

The Big Bang Bing 🌌

Complete with new hairdo and a fresh wardrobe, Bing is staging its mega-comeback, and it's happening faster than we expected:

Leaked pictures on Twitter picture a new and improved AI-powered Bing search engine that can answer any question ... with citations.

It's a step up from its younger cousin ChatGPT, whose prodigious capabilities have come under fire for failing to cite its sources (and worse, outright making them up).

Bottom line: Microsoft's new Bing launch could come with a bang. We'll keep our eyes peeled for any new announcements this week.

Smarter Than a Robot? 🤖

Introducing ... Social Media IQ Power Rankings.

We asked ChatGPT to rate the IQs of major social media platforms.

Then we asked Leo to do the same (don't get us started on how long it took to teach a cat numbers...)

Which list came from which?

A/ The internet-rabbithole list:

Reddit (115)
YouTube (110)
Twitter (105)
Facebook (100)
Instagram (95)
Snapchat (90)
TikTok (85)

B/ The Mr. Beast lovers list:

YouTube (115)
Twitter (110)
Facebook (105)
Reddit (100)
TikTok (95)
Instagram (90)
Snapchat (85)

Around the Horn 🦄

US investment firms funded 37% of China’s AI sector from 2015 to 2021.
An SF-based AI ‘virtual friend chatbot’ is being ordered by Italian watchdogs to halt all data processing as it “poses risks to minors.”
An overview of how AI text-to-audio generation might impact the music industry.
AI technology is helping rediscover lost work from Spain’s Golden Age writers.

Artificial OR Intelligent? 🗣

You tell us:

Get answers about any PDF you upload with Humata (ChatGPT for your files). Some ideas: textbooks, support manuals, insurance policies, work reports.
Quora releases Poe, a new chatbot experience for Q&A.
Pal: A custom AI assistant to help you manage your website.
Build a chatbot to be trained on PDFs or text you upload to it with Chatbase.
AI that lets you create custom beautiful icons for your business (IconifyAI).
Design Sci-fi-like images with an autocomplete interface that helps you write prompts faster.
A tool that will replace any photo of your ex with an item of your choice.
An AI song recommendation engine trained on 120 million songs from iTunes.
A free encyclopedia for all AI tools, podcasts, podcasts, and newsletters.
A tool that can distinguish the artists whose work contributed to generative AI pieces.

DM me links on Twitter: @nonmayorpete

Are you new to all this AI stuff? Here's The 3-Minute Guide to Slaying Your Dinner Convo About AI to get you up to speed. Or at least smart enough to impress your family.