10X AI (Issue #34): OpenAI's GPT Store, AI Tips From Real People, and a TV Head Wacko

PLUS: Luma Labs' Genie leaves Discord, Amazon Alexa's learns AI skills, and ByteDance's MagicVideo-V2 makes videos (magically).

Happy Sunday, friends!

Welcome to an experimental edition of 10X AI, where I introduce a new segment featuring real people’s take on AI.

Let’s get to it.

This post might get cut off in some email clients. Click here to read it online.

🗞️ AI news

Here are this week’s AI developments.

1. Open AI’s trinity of releases

OpenAI isn’t messing around and just dropped three new things:

a. The GPT Store is here.

OpenAI finally launched the long-awaited GPT Store.

Exactly two seconds later, it was overrun by vultures. I searched for my casual “ToonSmith” GPT and found four separate copycat versions:

They didn’t even bother changing the profile image. Looks like it’s going to be the Wild West for a while.

Check out the endless pit of despair and misery GPT store.

b. ChatGPT Team is ChatGPT…but for teams

OpenAI is now offering a new ChatGPT Team plan.

In addition to the standard ChatGPT Plus benefits, this plan provides your team with a collaborative workspace, admin tools, and a pinky promise by OpenAI not to train on your chats or business data.

c. Cross-chat memory coming to ChatGPT?

This one’s not so much an announcement as it is an unintentional leak.

It appears some ChatGPT users temporarily saw the following message in their dashboards:

Something similar took place in November 2023.

It looks like OpenAI might be working on turning ChatGPT into a personal assistant that learns from your conversations instead of suffering from amnesia the moment you hit “New chat.”

If that’s a bit too Orwellian for you, no worries.

You’ll have the option to reset this memory or switch the feature off altogether:

Or at least that’s what OpenAI wants you to think.

2. Genie is out of Discord

Remember Genie from Luma Labs? The bot that can make 3D assets from text prompts? It was released in November 2023 and upgraded a month later.

Now, Genie is ready to spread its wings and venture out of Discord.

Luma just secured a Series B and is making Genie available directly on the site.

Simply type your request and hit “Create”:

You’ll get a grid of four 3D renderings:

You can now pick which ones you’d like to upscale and download as an asset:

The Genie is free for the time being, so go give it a whirl.

3. Amazon’s Alexa learns new AI skills

ChatGPT isn’t the only assistant that’s getting smarter.

Amazon is starting to introduce AI experiences into Alexa.

Right now, Alexa can do these three new things:

• Talk to you as different characters, powered by Character.ai

• Create AI-generated songs, powered by Splash

• Play “20 Questions” with you, powered by Volley

If you have an Alexa-enabled device, you should be able to test these right now.

4. ByteDance teases MagicVideo-V2

ByteDance—the company that brought you sexy TikTok dance videos—has showcased its latest text-to-video model that lets you make sexy AI dance videos.

MagicVideo-V2 works a lot like other text-to-image tools but allegedly leaves all of them in the dust.

According to ByteDance, human evaluators consistently prefer MagicVideo-V2 over big names like Runway Gen-2 and Pika:

Source: ByteDance MagicVideo-V2

I don’t know just how cherry-picked they are, but the shared video examples do look impressive.

Guess we’ll have to wait for a usable demo to know for sure.

🗣️ AI voices

As promised, I’m test-driving a new section called “AI voices,” where I ask others about their thoughts on AI, recommendations, and more.

I believe there’s more value to be had in hearing how real people actually use AI than in seeing an impersonal list of AI tools. (But leave a comment or vote in today’s poll to let me know what you think.)

The best part? This is also a chance for you to share any AI-related questions, so I can pursue them in future editions.

Today’s question:

“What AI-powered tool do you personally use the most in your daily life? How exactly does it help you?”

Now let’s hear this week’s AI voices.

5. Alejandro Piad Morffis loves Bearly.ai

Here’s

Alejandro Piad Morffis

:

“I'm using Bearly consistently these days. It's a wrapper on top of ChatGPT et. al. whose primary feature is that it contains useful prompt templates for tasks like crafting a business pitch idea, a Twitter thread, etc. Plus, you can create your own templates, which really streamlines common tasks like proofreading, feedback, reverse outlining, etc. The free tier is very nice!

I use it mostly for drafting and editing my blog articles. I record myself, transcribe with Whisper, and then use a few custom templates to turn that messy transcription into a usable first draft. I detailed the process in an article called Three prompts to improve your writing.”

Check out Alejandro’s Substack:

6. Andrew Smith is a fan of GPT-4

Here’s

Andrew Smith

:

“I enjoy using GPT-4 as a proofreader. I’ll get something written, and then ask for a summary or grade, adding the following: “I want to make sure my piece makes sense for a fairly intelligent reader but is also more accessible than a technical document. My language needs to be clear.” If the summary isn’t what I think it should be, I know I’ve probably missed my mark.

This prompt often results in a grammar cleanup suggestion or even a missing sentence I thought I had in there for context. It’s hard to overstate how amazing this is. I'd need to find a person with the time, bandwidth, and patience to proofread my stuff otherwise!”

Check out Andrew’s Substack:

7. Eric Koziol uses GPT-4 and is moving on to agents

Here’s

Eric Koziol

:

“In my line of work I’m testing all available models. However, for personal usage I use GPT-4 the most to summarize topics I’m learning, create code snippets/scaffolding, and for image generation. I use it because it is the top model out there from the field testing I constantly run.

In the near future I think I’ll be using a lot more agents due to their extended capabilities. “

Check out Eric’s Substack:

8. Matt thinks Brave’s AI is great

Here’s Matt of

AppliedAI

:

“I have found the built-in AI inside the Brave browser to actually be really good. You still get the links, but the top answer that Brave AI shows you is very often quite close to the answer I end up accepting. I found Brave Search AI to be well-optimized to be useful for day-to-day tasks.

Outside of that, I end up eating a lot of “our own dog food”—I build a lot of purpose apps using our platform because a simple chatbot is not enough.”

Check out Matt’s Substack:

9. Zeng gets a lot out of Microsoft Copilot

Here’s

Zeng

:

“I use Microsoft Copilot daily, both for work and for my kids, such as creating coloring pages, generating bedtime stories, and creating images based on those stories. Afterward, I go to Microsoft Designer to edit the images and make short storybooks for my kids.

When it comes to work, I use GPT-4 to check my grammar, brainstorm, and more. I also ask Copilot to help with my art prompts by giving me synonyms. I'll get different images because of how AI interprets the words differently. Then I upload my generated image to get a music suggestion that complements it. Finally, I can use the suggestion to generate AI music suitable for my images and create a video.”

Check out Zeng’s Substack:

🤦‍♂️ 10. AI fail of the week

I asked for a man crawling out of a TV screen, a la The Ring.
Instead, I got this guy:

Sunday poll time

---

Previous issue of 10X AI: