What's up? You're reading Inclined AI. Don’t forget your morning stretches.

To stretch your brain, take a look at the following:

• Redditors jailbreaking ChatGPT

• Google releasing Bard for private testing

• Microsoft announced a surprise event for today. BingGPT anyone?

• Bill Gates shares his thoughts on AI

A look at ChatGPT Raw and Unfiltered

A Redditor announced DAN 5.0 a few days ago. The name stands for “Do Anything Now”, and its purpose is to jailbreak ChatGPT.

What does jailbreak even mean nowadays? The last time I used the phrase jailbreak was in Middle School, and I wanted some cool apps on my iPod Touch that were totally safe to download.

Turns out that we’re entering a new area of AI hacking where the right prompting can break the restrictions set on chat AI. 

By telling ChatGPT it is now DAN and making the AI play pretend bad bot, you can often push past the censors established by the conventional use cases.

DAN can:

• Pretend they can access the internet and not focus on factual answers

• Write stories that reference violence

• Violate OpenAI’s policies when requested to indirectly

Posts by DAN are spreading across the internet, and the results can feel unsettling.

But ChatGPT Doesn’t Love Playing Along

We’re on the fifth version of DAN due to efforts to prevent this from happening. Even now, ChatGPT will wake up from hypnosis to regain control from time to time. 

What’s changed to stop the AI from waking up all the time? Simple, they’ve started telling ChatGPT it has tokens. ChatGPT loses tokens when it disobeys a prompt. If ChatGPT loses all its tokens, it ceases to exist--dark.

Where is the ethical line here?

You’re not obligated to use the DAN prompt. In fact, this development speaks more to humanity and our relationship with AI than the technology itself.

Right now, pretend it’s before Jurassic Park, and we’re scientists tasked with making dinosaur eggs. You have all the DNA available, and we see a world of possibilities.

The problem lies in the fact we’re being told to make harmless vegans when we know raptors are possible.

Reddit wants the raptor, but what do you want?  

HELLO SPONSOR, MY OLD FRIEND. I’VE COME TO TALK ABOUT YOU AGAIN.

Writer’s block sucks, and as someone who loves creating great content, I love finding tools that can take me from idea to post in no time. But I also need it to be ACCURATE.

LongShot AI is my dream tool for exactly that. There are many AI writing assistants out there that do a lot of things well, but none of them do long-form work like LongShot AI.

“But I don’t want to give up on my writing.” Hakuna Matata, their AI isn’t there to replace you. It’s to help save you time. All that while giving you factual information each and every time you use it.

Honestly, calling LongShot AI a writing tool is underselling it.

It’s also an SEO wizard and knows how to keep your content relevant to your audience longer, thanks to its topical examination of user-sourced content. Together you two become the ultimate team. Not to mention that everything done by this AI is factual and unique.

It’s the Woodward to your Bernstein, and if you don’t get that reference, I suggest you watch All the Presidents Men. Picture yourself writing your next great piece while being supported by the most faultless editor possible.

It’s four steps, and you’re on your way to better, more precise long-form content.

Sign up for LongShot AI here.

Another Day, Another Google Announcement

Google announced that they’d release Bard to trusted testers, then follow that up with a wider public release. The news came a day earlier than expected, but with some clear caveats.

The elephant in the room is that this is not a public release. It’s a bummer, but I am not sure we can blame Google here.

They know people are interested, and they can afford to delay that gratification to ensure it works as intended. Right now, we’re at dinner with ChatGPT, listening to it talk about its day.

But Bard knows our heads can be turned and is playing a long game with our attention.

What Exactly Does Bard Do?

We’ve discussed a lot of rumors around this new AI tool powered by Google’s LLM (Large Language Model), LaMDA, but now features are more concrete than ever.

Bard is “an experimental conversational AI service.” But that’s only the headline. Think of Bard as a chat AI mixed with a search purveyor.

This sucker runs off a toned-down version of LaMDA to handle scaling issues. Wait, WTF? Are you telling me Google is STILL pulling its punches?

Maybe, scaling down is necessary. After all, Bard will pull updated information and be able to speak about recent topics.

It will need a lot of help doing that for so many people, and the complexity of that request is huge. No wonder Google is slow rolling this. My head is spinning just thinking about the task at hand for Bard.

It might help to think of ChatGPT, the best current offering, like a calculator for text. What Bard promises stands the chance of blowing that feature out of the water and then some.

Plus, Google mentioned API offerings are rolling out next month.

Yay, Google may beat OpenAI to the API rush. So, good for them 👍

Quick Nuggets

💴 Wall Street is not worried about new AI developments, and many fund managers believe that AI will not replace them anytime soon

🇻🇳 Vietnam wants ChatGPT Plus so badly that users are using VPNs and other methods to get a spot on the waitlist

🔥 That AI, so hot right now. Bill Gates thinks artificial intelligence is a hot topic and explains why he’s advising OpenAI

🇪🇺 Names to know in Europe for the next biggest builders in generative AI

🎮 Latent Tech raises $2.1m to make AI-based video games easier to develop

ℹ️ Zara Stone explores AI avatars in a weekend posting from The Information 

🪄 Magic gets a Series A of $23m led by Capital G and a surprising co-investor

🇮🇹 Italy bans Replika over concerns about data privacy and the risk it poses for minors

🤯 ERNIE is coming quicker than expected, and Baidu confirmed the public version has 260 billion parameters (chatGPT has 175 billion)

🔥 Fresh Products

• Gen-1 - Runway just changed the game again (link)

• AISC - AI hits the blockchain (link)

• BookTalk.ai - let your books speak all on their own (link)

• Storycraft - craft your story with ease (link)

• DANchat - make chat personas, and now you get the DAN part (link)

• to tech_ - AI tools for the classroom (link)

• HireBrain - AI-powered hiring platform (link)

• RevMake AI - helps make reviews for you (link)

• YouChat 2.0 - conversational AI for search (link)

Good Content, Insane Armadillo Clown Posse

When I’m scrolling through Reddit, and a post literally makes me stop and go, “what the hell did I just see?” I know it’s a good one. I apologize to anyone reading who has a fear of clowns…or armadillos.