OpenAI & Shutterstock | Shopify Sidekick| xAI

AI Daily | 7.12.23

Welcome to a brand new episode of AI Daily, where we explore the world of artificial intelligence and its impact on businesses. Today, we delve into three transformative stories.

1️⃣ OpenAI & Shutterstock

  • OpenAI and Shutterstock are expanding their partnership, providing training data for generative AI and potentially exploring the creation of Shutterstock's own generative AI, setting a potential trend for future partnerships.

  • The partnership seeks to proactively address potential legal concerns around content usage, shifting the responsibility of AI's guidelines and risk to Shutterstock, which may face repercussions if AI model training based on its data is contested.

  • Despite potential legal challenges, this partnership is seen as beneficial for Shutterstock, offering them access to generative tools and possible remuneration for their data usage, although the specifics of the payment model are yet unclear.

2️⃣ Shopify Sidekick

  • Shopify has launched "Sidekick", an AI tool within their platform, which helps entrepreneurs with tasks like changing images, adding text to images, and answering business-related questions.

  • The new tool streamlines the Shopify user experience by automating adjustments, such as changing header pictures, color themes, or adding new product banners, replacing manual fine-tuning with AI-assisted operations.

  • Sidekick's ability to handle abstract questions provides a valuable tool for e-commerce beginners, while experienced users may find it more gimmicky; however, it could potentially reduce costs associated with data analysis or consulting services.

3️⃣ xAI

  • Elon Musk's new project, xAI, comprises top minds from companies like Google and DeepMind, aiming to use AI to understand the universe, possibly launching a competitor to OpenAI.

  • The team plans to work closely with Twitter and Tesla, harnessing Twitter's extensive data and Tesla's advanced AI and multimodal work to create innovative multimodal models.

  • Rather than focusing on current AI tasks, xAI might aim at fundamental questions, like understanding the physical nature of the universe, potentially aiding research efforts at Twitter, Tesla, and beyond.

🔗 Episode Links

Connect With Us:

Follow us on Threads

Subscribe to our Substack

Follow us on Twitter:


Transcript:

Farb: Good morning and welcome to AI Daily. We got three exciting business stories for you here today, but, uh, since everything is ai, anything business related is ai and, and these are not, uh, these are not small stories. This is some pretty exciting stuff. Our first one is about the partnership between Open AI and Shutterstock.

I believe they're expanding their partnership here to. Essentially provide training data for generative ais. Uh, possibly Shutterstock creates their own generative ai. Who knows? Maybe they're just gonna apply the, uh, supply the content to open AI to do the work. This is a pretty exciting stuff here, and you can probably see a lot more of these types of partnerships happening in the future.

Uh, there is this whole tension between can you use this content without a partnership or do you need the partnership? Is this fair use? Is it not? Lot of open questions, but the smart folks at OpenAI like to get ahead of these stories, and I think this is another great example of it. Ethan, what are your Thoughts?

Ethan: Yeah, I've posed the question before, you know, are we gonna see in a couple years from now some court case come down that's gonna go retract these models? You're gonna have to retrain them, show the data you're getting for, so just like you said, open AI's getting ahead of the curve here. Um, they're pretty much just moving the risk to Shutterstock at the end of the day.

So they're gonna get to train all these models and the kind of guidelines around what AI's gonna look like is now gonna fall on Shutterstock for saying, yep, you can use all of our content to train these. We have. You know, a creator payouts, um, of sorts. So they're moving risk over to Shutterstock and they're getting ahead of the curve.

And yeah, we'll see how it plays out for the rest of the industries. You know, I think that question's always been interesting to me from a legal side. Are we gonna see these models have to get retrained some time? Are we gonna see people diving into the data behind it? How is this actually gonna build up when we start seeing more and more economic effects from it?

So, pretty cool stuff.

Farb: What do you think conner?

Conner: Yeah, I mean this, this is good news for Shutterstock. Also, if you look at any open source or even many closed source models for text to image, you'll see Shutterstock like captions in the bottom left, and for text to video, you'll see Shutterstock captions all over it.

So this data, this data's being used either way and now Lease Shutterstock is something to gain from it. Apparently they're working with open AI on some generative tools, and they're gonna get like first access

Farb: to it. So maybe Shutterstock poisoned all the LLMs and, and stuff to, you know, put Shutterstock, uh, watermarks on all, all the photos, you know, we'll ne we'll never be able to prove it.

Just kidding, Shutterstock, don't, don't send me a cease and desist or whatever you have l lawyers do. I think we're gonna see more and more of this. It, it's cool to see folks working together on, on, on things like this. And, you know, I was a, I'm curious to see how people are really gonna get paid for this.

Uh, you know, because how do you prove, you know, how might, is it just like, okay, well if you had a photo here, you get a penny because we used your photo. I, I don't think it's gonna be a per use case. It's not like Spotify where you get paid as people listen, you know? Uh, so it'll be interesting to see how, how this actually works out for folks.

Uh, but you gotta do something. Uh, and it looks like they're doing exactly that. Uh, onto our next story. Uh, the fine folks at Shopify, you've probably heard of them. Uh, Seems like about half the world's businesses are, are running on it. Uh, they're launching sidekick, uh, as they say hero. Entrepreneurs are heroes in their world and every hero needs a sidekick.

I think it's a beautiful angle and it seemed pretty powerful. I'm not sure if it's there on day one, but basically, You can talk to this AI inside your Shopify store and literally have it do things for you. You know, you can have it answer questions for you, but I think their goal is to actually have it do things for you, you know, maybe, uh, change the images that are in your store or add some text on the images that are in the, in your, in your store.

What do you think, Connor? What else did you get out this?

Conner: Yeah, I mean, launching a Shopify store is of course, already incredibly easy, and now for all your business ideas that aren't directly related to your store, you have an assistant that knows all that. But also as you said, like the main thing here I think they're going after is being able to say like, Hey, can I change the header picture of my site?

Hey, can I change the color theme of my site? Hey, I have this new set of products, of these new T-shirts. Let me add that as a new banner to people to buy. And it's, instead of going in and doing all the little editing and doing all the fine tuning, the AI will figure it out for you. And. It's another huge jump of what you can do with Shopify.

Farb: Ethan, you have a lot of friends in the e-commerce space. How do you think they're gonna take this? Or what are your thoughts?

Ethan: Yeah, I'd say two things. You know, there's a lot of like startups who have been working on features like this. Um, and now this has plopped right on the stop Shopify header, so now it's for everyone.

Now it kind of eats away at all that. One of the features I thought was pretty cool, they showed off in their demo, was actually asking these more abstract questions, right? So this might be pretty simple, but one of the ones they showed was, Hey, why are my sales down? And they're like, well, you're a snowboarding shop and it's summertime, so people aren't really skiing or snowboarding right now.

So it's kind of like abstract questions for people jumping into e-commerce are really interesting. You know, their whole thing was about building up the entrepreneur. For the more seasoned person on Shopify, et cetera. I think it's a little gimmicky right now, but it may help them kind of reduce some costs for contractors, you know, doing these data analytics or some trends or, oh, I don't know how to make the new charts in Shopify.

I'm just gonna ask the AI instead of paying some consultants. So a lot of people have been working on this, but like I said, now it's right embedded in Shopify from them. So pretty cool stuff. But Gimicky ish for now.

Conner: I think that's a great way to say it. If you're a seasoned shop Shopify user, it probably won't be much help to you.

But if you're starting a store for the first time, if you don't know how to do e-commerce, if you don't know how to edit your Shopify site, it's kind of perfect. It's a sidekick. Yeah.

Farb: I think, uh, maybe you can tell the sidekick to generate some snow so people can go and snowboard during the, during the summer.

I think one really interesting point that. I think he made at least a couple of times in the video, and it's along the lines of something I've thought a lot about over the years, which is that. You know, more and more because of technology, entrepreneurs are essentially becoming artists. And what I mean by that is, you know, most art is created by one person at a time.

Sure. You collaborate and there's plenty of great examples of art where there is collaboration, but with art there's usually someone with the strong vision and a lot of times it is just a solo artist doing everything themselves. And they kind of made this point a couple of times where. You won't, this thing is not going to challenge you on your vision.

It's not going to push back on what you wanna do. It's gonna help you get it done. And it seems to me like in not very long, we really will have even more. I mean, we have plenty of solo entrepreneurs out there in the world making tons of money. Um, But I think that'll just become more and more the case where you can just get things done on your own and the sort of merging of artist and entrepreneur is just gonna con continue until it's almost a circle of one,

Ethan: as I like to say, artisan.

Farb: Artisan. Yeah, I like that. I like that. You shouldn't trademark that. The, uh, the, the final story. Have your sidekick do it. The final story is xAI. If you haven't heard of X, it's, I think now the parent company of Twitter or, or, or something like, that's kind of tough to keep these trade. I think Twitter is part of x and x.

AI is not under X, but it's related to x. Uh, it's the efforts of Elon and, and a few folks to. Use artificial artificial intelligence to understand the universe as they say. Uh, Ethan, what do you think they mean by understand the universe?

Ethan: Um, you know, I think we're gonna see another LLM out of them.

I think we're gonna see some new novel research out of them. I think the big story here for me is a lot of these people came from DeepMind. Uh, a lot of people came from Google, so you got this complete brain drain right now over at Google and DeepMind. You got more competition in this space from, as everyone knows Elon.

So yeah, you're seeing a brain drain from the other companies. You're seeing more people jumping in the AI space. I'm not sure what they're gonna launch, but I think we're gonna see. You know, a similar-ish competitor to OpenAI. Hopefully some novel research.

Farb: Okay. L l m you're thinking, Connor, what do you, what do you think?

L l m something else? Somewhere in little This little that, yeah,

Conner: they, they're saying they're gonna work closely with X or Twitter and Tesla. So, l l m, I'm assuming probably some spatial stuff with Tesla. Um, But yeah, DeepMind opening ai, Google Research, Microsoft Research. These are top tier minds, and it's the top tier team they put together.

Ethan:  I'll say one more thing too. Twitter has a ton of fantastic data for something like this. You know, videos annotated with comments of a tweet, text annotated with an image. So in creating these new multimodal models, they actually might have a pretty interesting advantage here. That can help out. You know, the Tesla bought robotics that could help out some of Tesla's multimodal work they've been doing for a decade now on the car.

So being able to bring that to these kind of more general intelligences with Twitter's data combined with the intelligence of Tesla and a great team from Google Brain, et cetera. I think they're onto something, you know?

Farb: Go ahead, con.

Conner: No, I was gonna say like Elon's done a lot of work lately of course, to try to try to protect Twitter's data from like login, pay, login walls to bunch of like bot protections to even stopping like embeddings of tweets when you send 'em over iMessage.

So clearly Twitter's data is what he knows is valuable and why he probably bought it and he knows that's what he wants to use it for when it comes to X ai.

Ethan: Yeah, we thought it was the threat of threads, but it actually might have been open ai. They were trying to stop.

Farb: The, uh, it's possible. Yeah. I'm gonna go out a little bit on a limb here and maybe try and synthesize some of what, what I'm seeing.

You know, Elon likes to go big. He's not usually one to sort of be, I'm going to copy, uh, what someone else does. He, I likes to. Go way far ahead. You know, leadership is creating an unattainable ideal. Uh, if the ideal is attainable, it's, you don't usually get as far as if the idea is sort of almost unattainable.

And so I think by positioning this as trying to make sense of the universe, and this is again me synthesizing a few things I've thought and seen is, uh, I think one of the folks on the team tweeted about how, uh, this is going to have a lot to do with math. I think they want to use AI to try and actually tackle some more fundamental questions because AI is not great at doing insane types of mathematics right now.

So, you know, I don't think Elon wants to make a better l l M for, you know, as a personal assistant. I think they want to. Shoot for understanding the physical nature of the universe more closely and assume that there will be plenty of business opportunities that come from that. Like I think it maybe both of you mentioned, uh, you know, all of the computer vision stuff they're doing at Tesla is truly around trying to understand how the physical universe works.

Uh, obviously mathematics is hugely important there. Uh, they have optimists, which is a robot that has to move through physical space. So my sense is that they're going to take some, you know, pretty deep swings at what they think al artificial intelligence can do, uh, beyond what it's done. We'll, we'll see.

It's, uh, even if it's not that, uh, I think he's right to, uh, set a vision that is, let's understand the universe fundamentally, uh, with AI as opposed to understand, what's the next word that should appear in this sentence?

Conner: Yeah, like their entire team's in a pretty wide array of research. And from the angle of understand the universe, I wouldn't be surprised if it becomes something like AI that helps them research further for all the AI they're doing at Twitter for all the AI they're doing at Tesla and anyone else they work with.

Farb: So. Very cool. So, uh, what, what, what are y'all seeing here for, uh, for my part? Uh, a friend of mine shared a cool Disney AI trick where Disney is trying to build de aging into. They're tech stack so they can de-age artists. When he texted it to me, he said something to the effect of, uh, you know, I heard half of Hollywood was working on trying to de-age Harrison Ford, poor Harrison Ford, uh, uh, you know, And obviously I think his point was, it's a, it's a big task if, you know half of Hollywood was working on it.

How can, how can Disney do that? Obviously super valuable for somebody like Disney. Uh, I thought that was pretty cool. We'll try to link to that here in the, the notes. What have you seen conner?

Conner: I think Harrison Ford looks fine just the way he is personally. Yeah, I do too. But yeah. So objaverse-XL.

Um, basically just a data set of 10 million labeled 3D objects. Not super high fidelity, but it is interesting. It is nice to have another data set of well labeled cleanly cut 3D objects. No bad data in there or anything, so Very cool. Very cool.

Farb: What about you Ethan?

Ethan: Uh, I saw this, um, it came from a meme page, but it was also a paper that was only PR and caban, no data, but they were claiming GPD four matches the top 1% of human thinkers on a standard test for creativity.

Wow. I'm gonna link it below. I, I have many, many comments, but I don't even want to get into it. So read it below and see for yourself.

Farb: Wow. Okay. Well maybe we can get you to get into it on another episode. Here. Everyone go do your homework. So Ethan will tell us what he thinks about it, the, uh, Teacher will appear when the student is ready, as they say.

Thank you for joining us on another episode of, uh, AI Daily. We are just about a hun 500 subs. We're just about at 50 videos record, 50 episodes recorded. We can't thank you all enough for your support, uh, and the incredibly encouraging words that you've been leaving in the YouTube comments. We read every single one of them and try to reply to every single one of them.

So thank you very much and we'll see you again on another episode. See you again.

0 Comments
Authors
AI Daily