Stability AI

The essential guide to a sector that’s now attracting one out of every four startup investment dollars.

AI-Innovators-2023-Stability AI

Stability AI is best known for its efforts in helping create the ultra-popular text-to-image generation system Stable Diffusion—where users can input a prompt and create an AI image of anything they want—but the startup is working on a host of other AI projects that are open models, meaning anyone can access them. This spring, Stability released two language models, StableLM and Stable Vicuna, as well as two additional imaging models, according to the company. It has also created a text-to-audio model called Stable Audio that it released in September. Stability has been an early and vocal champion of open-source AI, where people can access the entire model, all of its code and weights, as well as all of its outputs, for free and with almost no restrictions. Some big investors have gotten on board with the startup’s creed, as Stability has raised over $110 million, per PitchBook data, from the likes of Coatue Management and Lightspeed Venture Partners. Over 200,000 creators, developers, and researchers, as well as seven research hubs, use Stability's products, according to the company. However, Stability and its founder Emad Mostaque have also recently come under scrutiny: Mostaque reportedly has a history of exaggeration, and a Forbes investigation found that several of his assertions about his own background and Stability's customers were either false or inflated. Numerous senior executives and AI experts have reportedly left the company this year. However according to the company, overall headcount is up, and they increased their revenue base more than 10x this year. And the company got a boost in early November as Mostaque posted on X that “We closed strategic funding ourselves last month (announcements soon)." <a class="waffle-rich-text-link" href="https://www.bloomberg.com/news/articles/2023-11-09/stability-ai-gets-intel-backing-in-new-financing?srnd=premium-uk&amp;sref=7H70AtGu">Bloomberg</a> later reported that Stability had receieved a $50 million round of debt financing in a round led by Intel. But it also reported that both Coatue Management and Lightspeed had declined to provide the company further financing and given up theirs seats on the company's board following disagreements with Mostaque over the company's management and direction.

Fortune 50 AI Innovators

Get insights into Stability AI including business profile, stock information, news, and rankings from Fortune.

Stability AI Company Profile, Stock Price, News, Rankings

legacy

stability ai

Companies

Stability AI | Fortune

Stability AI Company Profile, Stock Price, News, Rankings | Fortune

Financial institutions are grappling with unique challenges in the age of AI, including keeping up with head-spinning innovation at legacy organizations designed to manage risk. How do these leading COOs in finance lean into long-term stability while staying agile in the face of change? You’ll hear about how these global institutions are using ground-breaking tech to modernize the customer experience, detect fraud, conduct cutting edge research, streamline processes and more. 

INDUSTRY DEEP DIVE: HOW FINANCE COOs ARE LEADING THROUGH TECHNOLOGY’S RISKS & RETURNS

Conferences

Well, OK, as Diane said, AI has the potential to be transformative to the finance to the finance function. So I want to start by asking both of you. You, you, you're both in big financial institutions. Where does your institution, Jack and we'll start with you. Where are you on your AI? Path Thank you. So let me start by just providing a brief background. RBC Royal Bank of Canada, uh, is one of the largest global financial institutions. More than 100,000 employees as part of our AI journey, happy to say that we started our MLAI journey almost a decade ago and where we are just recently earlier this year, we shared in our investor day and made a very bold, ambitious goal that by the end of 2027. We will generate $700 million to a $1 billion of enterprise value through AI-based initiatives. And the reason we were able to make such Uh, bold back is based on how we have, uh, responsibly invested in technology over the period of last decade starting from Whether data analytics, API, private cloud, public cloud journey, uh, machine learning, AI, and now what we all are really hyped about generative AI. Bye. Yeah, thank you, Rashmi. Same thing. Yeah, so you know, TIA is primarily a retirement company. We oversee about $1.4 trillion in assets. I'll walk you a little bit through our AI journey. We are a very mission oriented company. So it was important for us to ensure that when we innovate, we innovate in a responsible manner. So one of the first things we actually did was establish a responsible AI policy which tells our employees the dos and don'ts of AI essentially. We also then created a center of excellence for AI. We have 3 major business lines. We have our retirement business, Novien, our asset management arm and wealth management, of which I'm a part. Each of these businesses identifies the use cases based on the, you know, their work process and what they need, and they work with the center where we govern those use cases. We train employees, we monitor the usage of AI, we deck the right capacity to it. So really setting up, I would say first and foremost the controls, the governance was important. The second thing we wanted to do was obviously create uh the ability to build AI solutions. So we set up our own in-house proprietary AI platform called TIAA Gate. Um, it's in-house, but I will tell you our strategy in this space is to be hybrid, to be hybrid hybrid, yeah, because you know, the best solutions are not going to always be in-house. So the approach we're taking is to say. We'll leverage third party solutions with providers we already work with, for example, you know, we may work with Salesforce and use one of their agents, and then where the solutions are unique, for example, our lifetime income products are pretty unique in the industry. So for those solutions, an AI option is best done on our own proprietary platform. So really our strategy is going to be hybrid, um, and I would characterize the evolution of AI is. We set the tools. We've got the governance, we've got the controls. We've started the training, upskilling. We've experimented with use cases. I think the next phase is really beginning to do the use cases at far more scale. Yeah, yeah, hybrid for you also at RBC. Yeah, it's very, very much aligned and I think as I, as I was saying, right, as part of the last 10 years of journey. All the lessons we learned from the previous technology that came and we all were talking in various conferences, right? We learned a lot. We harvested all those lessons learned, whether it's related to data privacy, the security aspect of it, making sure we understand the ethical nature of uh using the technology, being responsible about it. So similar to what Rashmi is saying, right, we, we applied that as part of our cloud journey because in order for us to leverage public cloud, we had to go through the same rigor on top with respect to data privacy and security. We use this foundation to build a similar level of governance for our AI, and now we are taking that for JNAI, right? And as part of that AI JN AI journey, we also have, uh, responsible AI principles that we have established and it's publicly available. So if anyone wants to check it out, you can, you can access it. It's a publicly available information for RBC and as part of it, we have also invested and created Borealis AI Institute. Which is basically, uh, employing 100+ PhDs, world's best, I would say AI scientists, and 800+ machine learning AI developers and data engineers. Uh, and, and the reason for that is that has helped us. Understand not just the business applications and the power of AI, but also understand how we can Leverage all the investment we have made over the period of the last many years using technology because at the end of the day with the recall of AI GAI or agent T AI, it has to tap into your ecosystems of the data which is your primary asset that you are hosting uh related to your customers across the line of business. Um, and then what we have done is we have created a very, uh, detailed development and governance framework. To allow various teams in a large RBC organization to understand, learn, design, develop, and test AI and AI model in a very responsible manner, and there is a very clear guidance and governance when they can make a decision how and when to take this to the production, so there is a very well laid out structure. Got it. So let's get very specific this time. What are the current use cases that you are working on now? Yeah, you know, I would categorize the use cases we typically see in wealth management in three different patterns is how I describe them. Uh, content summarization, content generation, and large scale data analysis for the purpose of insights. There's also core generation, by the way, but that's far more, I think, in the tech space. So these are the patterns we see. The impact of a lot of these use cases is AI as a co-pilot and AI not as an autopilot necessarily for our employees and advisors. So I'll give you an example of a use case. We are working on a meeting preparation. A use case for our financial advisor. So when an adviser is meeting with their client, they often have to spend time preparing for that meeting, right? What were the last notes from the last interaction? Did the client call between the last meeting and this meeting, and did they have a service issue? Are they on track with their financial plan? Are their life events likely to happen? This takes time, so we are actually. Looking at a solution where we can use AI to create a summary for the adviser that saves them time and then frankly go the next step and say give them recommendations on the next best action to have a conversation with the client on and by the way, Jeff, we think there is tremendous value in this because there are studies that have said financial advisors are spending 40% to 50% of their time on non-client facing activities. And close to 15-20% of that on administrative tasks. So the opportunity is clearly there to get more productivity. Yeah, and a follow up there, what you're describing would be equally valuable to many other financial companies. Are you creating your own? Are you using some that at least have elements that are being used elsewhere? We are definitely using creating some new ones and then leveraging others that we see can be applied enterprise-wide. So an example of another use case that's very common in the investment spaces you want to be able to summarize the investment researcher portfolio manager. uses to make investment decisions. By the way, our asset management firm Nve has to do the same thing, as do our wealth management investment advisors. We don't need two separate use cases, and that's where that center comes into play so that we can govern and leverage core capabilities. Yes, use cases now at RBC. Yeah, so. The way we look at AI and AI is two ways, I would say. One, how we can provide, uh, services, products and services we are offering to our. Our clients and the experience we are giving to our employees, how we can do this thing better. And secondly products and services or the experience that we wanted to offer to our clients and employees that were not possible that something now AI and GNAI is making it possible now. So these are the two different buckets. So when we talk about use cases for the period of the last 57 years, we have, uh, Made AI machine learning AI based products and services available to our clients across different line of business. Uh, if I talk about capital markets, we have Eden, uh, that's a brand name, uh, that is a MLAI based model that helps with equity trading, and it allows our client to have a better efficiency and Uh, quality of their trades. Now we are taking to the next level and it's all helping us with the research and insights and as, as Rami mentioned, right, enabling our frontline, uh, team basically to focus more on engaging with clients and having a quality, uh, conversation versus doing a lot of backend research work by themselves. If I talk about from a retail perspective, uh, through our RBC mobile app, uh, we have made available know me insight. Uh, what it does is basically it looks at all your financial cash flow transactions and allows at an individual level to avoid you to go below a certain balance and it automatically if you if you're given permission to move money from savings to checking because you have set up autopay. And it knows that there is a bill coming and it's roughly about this amount. So we don't want our clients to go through insufficient fees type situations. So we are leveraging technology to make our clients' life easier and same way to small market customers, they are able to look at their cash flow and understand and predict and help them properly run their business and and focus on their business rather than worrying about the, the financial cash flow aspect of it. If you, if you go into the broader enterprise side, same way there are a lot of use cases, but the foundational is. What we are most proud of is this BorealS AI and and 800 to 900 top of the best of the best world's talent have created a proprietary large transaction model, we call it atom. It is Running across 1 billion events that we generate every day, and it was trained using 65 billion client transactions. So we are, we are sitting on this intelligence that is allowing us to unleash the power to help our employees better serve our clients and our clients receive the product and service. That they were hoping but we were not able to provide so far the quality and experience that they were hoping for. Yeah, super exciting. There's a human element in this as in everything, it does seem to me that whenever there is something new like this, especially profoundly new, uh, and I think especially so in the finance function, there will be resistance to change, right? It's inevitable. How do you respond? So you know, I would say. Given the nature of the use cases we're seeing AI as a co-pilot, it, it really means humans who use AI will replace humans who don't. And so getting the human element right and getting them ready to use AI, I think is to your point, the biggest focus area now, and I think you need 3 things for that. You clearly need to give them access to tools. You need to give them the skill set. Right. Um, and then you need to, frankly, in our experience involve them in the creation of the solution. So co-creation is a big lever. um, so to give our folks access in March, we took our proprietary AI platform and created an associate version of it where people can now and we're rolling out access throughout this year so our employees can go in and use AI for their day to day tasks. Whether it's editing documents, writing emails, creating meeting notes from a Zoom session, you know, whatever it is, um, but it gets them comfortable from an upskilling perspective, we have focused a lot on, it's an interesting concept. We've established guilds. There are about 9 different guilds. These are informal networks in the company where employees come together, learn, and experiment. We've got over 5000 employees already members of the guilds. And then I will tell you co-creation we had, you know, we have estate planning attorneys who create, uh, give estate planning council to our clients. They spend a lot of time going through the estate planning documents and they're testing two different AI solutions, giving us a lot of feedback, and their involvement in it has been, I think, instrumental in driving adoption, you know, for both of you. There is also a leadership element here. I mean, you are at the heads of large organizations within your organizations. What's the leadership lessons you've learned so far in this very uh, profound change, so. All the, I would say leadership lessons that we all have aggregated over the period of many years, 1020 years when new technology came, I would say all that applies one or the other way, right, because I, I think your question is specific to AI AI, so I'm going to switch and say, assuming that we are not chasing technology for the sake of technology, we are applying business plans, we are being responsible about it. We are thoughtful about it. We are thinking about hybrid and not just going one way and then learning later, right, that this was the wrong path. So all those given what AI and GAI, if I, if I make it specific leadership lessons, is one is around. being in a highly regulated industry as a global FI. How do you, how do you create an ecosystem? Where the individual end user, the client, customers. is also trying to figure it out what does AI AI means for them. So we have partnered with a company called Kohre. Which is basically, uh, helping us create uh a financial services GNAI framework platform to allow, uh, creating a platform where our end user and client can develop their product and applications using GNAI. Why? Because we want to make sure that there is an independent party, uh, who we are partnering with who is protecting the security and privacy aspect of the data. We are sharing the relevant information, uh, and the, and the learning that we have. And the second part is as part of the broader journey. First, you have to win hearts and minds of your employee when you are going through this change. But same thing applies to your client, same thing applies to your community as well. So we have established respect AI. As a portal, as a framework where we are openly sharing tutorials, framework, knowledge article, that is also something available in public. So if anyone wants to to check it out, to help yourself, help your company, do that. And the reason is this is where we are. We cannot just take this new technology and say. Smart enough, we understand, we figured it out. Let's just use it. We need to think about the overall ecosystem. Great, Rami. So I would, I would echo, don't do tech for the sake of tech. I would say focus on the foundational. I've called it fix the leak. Don't just patch the drywall, do that. The other thing I do want to say is the skills of the future are going to be different. There's going to be a premium on certain skill sets, so speed or structure. Insight over title, judgment over certainty and frankly clean slate thinking or continuous improvement. So we as leaders need to ensure we are evaluating for these skills, hiring for them, performance managing for them. That's where the future is. Uh, that's a great note on which to end. And so I want to thank Jakin and Rajmi so much, really good. Thank you. There you go.

Prem Akkaraju, Chief Executive Officer, Stability AI, Dana Settle, Co-founder and Managing Partner, Greycroft Moderator: Allie Garfinkle, Senior Reporter, Finance, Fortune; Co-chair, Fortune Brainstorm AI


Brainstorm AI 2024: A New Lens on Stability

Prem Akkaraju, Chief Executive Officer, Stability AI, Dana Settle, Co-founder and Managing Partner, Greycroft Moderator: Allie Garfinkle, Senior Reporter, Finance, Fortune; Co-chair, Fortune Brainstorm AI

Prem Akkaraju, Chief Executive Officer, Stability AI, Dana Settle, Co-founder and Managing Partner, Greycroft Moderator: Allie Garfinkle, Senior Reporter, Fortune; Co-chair, Fortune Brainstorm AI

Frame Dana. I am so excited to be here with you in part because this was known on my Bingo card for 2024 talking about a turnaround for this company less than eight months ago, stability was in a rapid decline. But here you are, you both stepped in and decided you really wanted to take it on why it wasn't on my bingo card either actually. But no one, no, eight or nine months. Yeah, it's no one's Bingo's Card. I think that um well, it, it was really because the, the, the technology and the team were too good and uh I've been partners with Sean Parker for over a decade and we have just did a very similar uh play at Weta Digital where we created an application layer for film and TV. We made Avatar one Avatar two big movies at that with that company. And we were after and that was we sold that company in November 21. And after that, we're like, ok, we should, we should probably be looking at the A I version of what we just did and how could we use A I to make film and TV, production gaming, as well as marketing, advertising, more efficient. And then this opportunity came about and I got heavily involved with Dana um on this like earlier this year. And um you know, this is where a lot of the outweighed returns live is when you're willing to go in and do the very hard work to kind of turn it around. And we did that very quickly and in opportunistic situations. And I think, um I'm, I'm fortunate to be a part of the uh the partnership now, the tram brick along with many other partners. Uh But, you know, like Prime, that is where the outsized, you know, returns live is actually in jumping into opportunities that, that other people don't always see. Um, and especially in an area like this, that's moving so quickly and where the technology here, like Prime said was just, it was, it was too good to let go. Dana. Dana bring the, sorry to interrupt the, the uh um she brings up a very good point. Like opportunities are things that only a handful of people see if we all saw them, they wouldn't be opportunities. And so that, that's kind of the definition and I think that we saw it and everything that we've done for this. How many months has it been now? It's been six months, six months, last six years. How long a time? Exactly. It's like 100 and 52 years in like real time. But, uh, we, we've seen extraordinary results in such a short period of time but, and it was really, it's, it's good to see that, you know, our assumptions really coming through and what was the financial state of the company when the two of you came in? What were we dealing with here? She's the, yeah, I mean, what are we looking at the financial state of the, they needed funding and, and that was, you know, where we had to really move quickly on analyzing the opportunity. And so we moved very quickly, but so deeply because we wanted to make sure that it was what we thought it was. And, you know, like Prime said, it's proven out to be, I mean, e everything we thought and more and I think in particular for us, it was really backing an incredible leadership team in Prime and Sean Parker to, you know, come in and, and lead what is essentially doing what they had done just a few years earlier in the South Space. I mean, you said they needed money. What for how bad was it was it because the reporting at the time, my colleague, Jessica Matthews and I spent a lot of time reporting on the story and the, the narrative was that it, it stability was a dust store. Well, I mean, look all A I companies require a, a lot of capital. I mean, that's no surprise. I think you can read it. Anywhere. Um But you know, I, I think what, what happens when you come in with a really incredible experienced management team, leadership team is they know how to get in. And this is prime, I mean, Prime really was working with the team um at stability even before we close the transaction and, and uh and, and helping to identify, you know, exactly where were the areas that they needed to focus on for this opportunity that we saw in visual media. I would say that people forget that Apple was 90 days from bankruptcy in 1997. Actually, I think they were probably closer to bankruptcy than actually, you know, this company was. And again, it all comes down to leadership and you saw what happens when leadership comes in with a very succinct plan, a product vision and the ability to, to attract capital obviously. But, you know, the clean up, we have now a clean balance sheet. We have no debt, we have nothing and to negotiate out of those deals or within those partnerships. It was no miracle by my part, even though I led the transaction. Give me an example of what we're, how much of a mirror. Well, I mean, it, it was, again, this wasn't on anybody's bingo card again. Well, I'll take, I'll actually give the credit. Well, I appreciate that. Thank you. But um while I led that effort, it was, we, we cut deals in the best way. That you cut deals, which is um alignment of interest. And everyone on the other side of the table that I had to deal with to clean it up. And then we got there was saw that this was essentially too important to the ecosystem to fail. Stable diffusion. Represented 80% of all the every single image that everyone in this audience saw in 2023. Just last year, 80% of them came from stable diffusion that's gigantic. And when you think about what that implies with a 5 million person active community that works on our models every single day that's growing every single day. That's a gigantic ecosystem that the infrastructure of this industry could not take a hit on. So I think it was part, it was, it was, it was the alignment of interest with those parties and um uh you know, and our determination to make sure that that was articulated. And, but I mean, the reason I say you should take a lot of credit for it is, I mean, really bringing everybody along at a time like that is, I mean, that is true, hardship and that is hard and it's in doing it too when, you know, I mean, frankly, you weren't even ceo the company. And so, I mean, I think when you think about how difficult it is to lead a company during, you know, tumultuous times and during super high growth and, and change, that's hard when you are the CEO but, but, but what about when you're actually just coming in and, and sort of coalescing, you know, all of the different groups from, you know, investors, employees to partners, et cetera. And I think that, you know, that's what gave me greater and greater conviction at every step along the way, was just seeing how prime actually executed through that. Well, controversial past leadership has left the company but a lot of the early investors in the company, including including Co two and Lightspeed, they left and then they came back. Yeah. How did that go? How did that happen? Is that incredible? I'm very, I'll be honest, I'm surprised that I think that says it all, you know, and I think they really re underwrote to this leadership team but also understood deeply the opportunity. Co two left the board, they didn't invest anymore. Same with Lightspeed, same with sound ventures. So not only did all of three of them reinvest into this round, um their pro rata that they were allowed on only the best we did it, we did it as a rights offering in, in, in some cases even more than their pro rata. And not only did they did that but coach who actually rejoined the board and is on our board again now too. So it was uh it is extraordinary because I think they saw that progress being made and kind of the positioning of re establishing the leadership position. The reason that everybody is interested in the company is because they were so successful. If they weren't successful, you wouldn't have these, like, no one would care. What are those conversations like that? What was the vision that you, that you sold them on? Well, again, it was alignment of interest, I think in those, you mean for the other investors, it was really well there, it was really my pre previous success. And yeah, yeah. And I was gonna say, I mean, I think one of the really exciting opportunities here that, that we saw is a new investor coming in was essentially um you know, where we think that there is great opportunity in general A I right now is, is really when you're sort of coupling the models with the application there. And I think these models were phenomenal are phenomenal continue to be and then having a, a leadership team that's so deeply throwing stuff around under understands the the do exactly heads up um is really an extraordinary opportunity. And I, and I think that's again why I mean for us as new investors coming in and, and we are based in Los Angeles, New York and Silicon Valley. But we, we really understand the media ecosystem and knew how deeply prime understood that and, and was connected in that ecosystem. That's really that, that transition because we were, you know, the first evolution of A I was really the foundation of the frontier models. And then it was quickly seeing this base, um because people always talk about the application layer and it is true that value is there. But being able to understand and fine tune models all the way into a core model is extremely important for you to set up success on the application layer. So I don't think, well, you know, some people believe, uh you know, foundational models are, is raced to the bottom. I I, you know, I uh I tend to agree as a stand alone. Um We're really that application, the applied research and application layer, which is a race to the top, but that's a race to the top that we can have an advantage winning when we understand the entire entire software uh stack all the way down to the chip. Now, I want to talk a little bit more about the board coach. He's back. You da are on the board. James Cameron is also on the board. Yeah. Who here has heard of James, I actually who doesn't know who James Cameron is? Avatar? Avatar? Yeah, just a just stream avatar. Um How did that happen? What does he bring to the board? So I worked. So when I was CEO of Weta Digital, we made Avatar two made every frame of that movie with, with uh James Cameron. And so um when I built the application layer at WETA and eventually sold it to unity, uh you know, he had a front row seat at everything that we were, I was doing and uh with me and the engineering team. And so when this came about, um I was in constant dialogue with, with uh with, with James. And, um and in fact, pulled him in early in the year before I joined the CEO and I said, hey, I think this actually could be. And he said what he said, he agreed. He said he agreed, you mean about that? This could be, this has the potential. Of course. Well, I don't think he would have invested or joined the board if he didn't, I assume. Yeah, that makes sense. That track. But even back then before you joined, before it was before this full turnaround happened. Yeah. Well, no, it was definitely, it would mean he agreed, it would be hilarious if he did join the board without thinking about that. No, no, no, no, no, no. Well, there was no board for him to join at the time, right? I wasn't even the CEO of the company. So what I mean was he agreed was, was the question that I posed to him, which was, hey, I think this is worth looking at and like, let's really dig deep into this. And like Dana was saying, we dug very, very, very deep into the talent pool. Like we didn't, we didn't invest and I didn't join the company until I met every single person at the company myself. Um and Sean and, you know, we dug very deep in the talent pool was so depth, was so deep. I was so surprised because there's a lot of also like in no offense, the media kind of, you know, over exaggerates in both directions many times. And then so II I was reading but she's laughing because it's true. Um But I was so shocked because this was because there were some departures and there's other, you know, spinoffs that came off of it. But we were so um so happy to see that not only did the depth of talent still stay there. And there was over 90 people within the core and applied research team, but they had two other things that were, that you that are beyond just their genius and they were dedicated and they were very passionate about stable diffusion. And I found, and I, I told this and just recently, um during all that, you know, that was what feels like years ago, um was months ago when I was meeting everybody and when we close the deal, there was somebody on the applied research team that said, we knew you were coming. This is even before the actual deal was announced and closed. And they said, and what's that? Yeah, I mean, it's an amazing statement to make meaning, not me, but like, you know, somebody was coming that was experienced leadership that could bring the vision that could bring the capital to the table and clean up the company because it was just, it was too important. Well, they might not have imagined that it was gonna be you, Sean Parker and James Cameron. But to be fair, who could have imagined that again? Everybody's surprised Bingo card. What is stability do to today? And what is the opportunity? It still it down for us here and now. So our vision is we're doing two fold is one is we're exploiting the commercial opportunities, significant commercial opportunities with our enterprise uh licensing and API revenues. And the second is we are now building the first, first full stack A I production pipeline for professional vi visual media content, professional video media content, defined as film and television, obviously gaming and marketing and advertising. What about you? The opportunity? That's where, where, where does, where does this land for you? What's the end game? It's, I mean, it's amazing because it really is sort of the thesis that we had laid out at the beginning, which is that, you know, there is a huge opportunity on the creative and production side of media to leverage A I in the creative community. And really, I mean, you know, if you think about it, we have this incredible product marketing genius and James Cameron on the board deeply involved and having that combined with this technology is so unique and being able to then apply it to those three verticals verticals that, you know, we at gray crop know very well. Um Having been involved with, you know, companies like Scope Le, uh which is a large gaming company which was sold uh last year, um, you know, early days of youtube with Maker Studios off the TV. So this is an area that we're in, you know, mountain on the advertising side. We just think there's such tremendous opportunity. Um Sorry. Yeah. No, no said perfectly, but I, I actually don't know how to do it any other way. Um I think it's pretty because I did it at Weta. That way, we were, there was no space between the art and the science. Uh We worked with some of the biggest directors in the world. We made the last 10 Marvel movies and, you know, obviously avatar one and two and we worked very, very closely with the directors. And uh that was such a great um experience for me because people ask me all the time after we sold, after I sold the, the pipeline to unity. It's like, how did you know what products to build? How did you who designed your product pipeline? I was like, it's easy. I just asked the directors what they wanted. And so it, and it, it's true. So it, it became the product strategy, the product road map, the go to market strategy, everything and also the Q A everything is sort of built into one. And uh the confusing thing I find is how am I supposed to build uh artist tools without an artist as a partner at the highest level of the company? So that's just it, it's my, it's my muscle memory. Now, let's imagine for the sake of argument, we're sitting here in a year, us and a couple and a couple of our nearest and dearest. What has stability achieved? What does stability look like a year from now? Like robust managed service on our um which is growing at multiple, multiple, the, the, the, the head of marketing told me I couldn't tell you how uh in front of everybody, how much we've grown. But it's been extraordinary. Exactly. I would like to, I would like to because it's a gigantic growth. I know. I know. I blame you. Yeah, exactly. I'll tell you in a year, I think, let me then on here you can ask her the um uh so she wouldn't let me talk, but it's, it's incredible. It's triple digit growth and the uh and so I think we're gonna continue to see that and I think that we're gonna be seeing um several breakthroughs in individual shot elements. We have a very different approach. We have a professional approach to content creation. We don't have one text to video model that has one output that combines like a gazillion, different workflows into one that's actually not how professional content is created, professional, created in shot elements and then they're composited together, by the way, this is where he's going to go really, really deep because he doesn't understand this thing at all. Models. Now, if you want the, we're sadly at time. But that's, that's, that's for, it's ok. I can go over. I mean, they all came and way more interesting. Thank you and Dana so much. That is all the time we have. I wish we could. Thank you so much.

Emad Mostaque, Chief Executive Officer, Stability.Ai

Brainstorm AI 2022: Silicon Valley’s Newest Gold Rush

Uncategorized

Brainstorm A.I. 2022: Generative A.I.: Silicon Valley’s Newest Gold Rush

Welcome back. I hope you all had a chance to stretch your legs and grab a coffee. And now let's continue with our main stage conversations. There is probably no hotter topic in A I these days than generative A I. It's an umbrella term for software that instead of just analyzing existing data can create original blog posts, Wikipedia entries, artistic images, video code, and more up next, a demo of the remarkable tech from Stability A I followed by a conversation with the firm's founder, Ahmad Mustaq. He'll lay out the economic and creative upside that will follow from making this open source free to use code available to anyone who wants it. Please welcome to the brainstorm A I stage, Ahmad Mustaq founder and CEO of Stability A I. We'll get to the demo in just a minute. But first of all, I want to ask you, why did you do this? And especially why did you do this as an open source project? I mean, potentially, you know, this could make a lot of money in the film entertainment industry. Why do it open source, fun and profit? I think I saw this technology and this technology classical A I was you built an air for one thing, this is there that can do anything. And I realized it would be like servers and databases which are almost all open source and it's all about scale and service customization. So I thought rather than wait five years to get there, let's just go there at once and allow it to spread everywhere which it has. So like for example, stable diffusion has just overtaken Bitcoin and Ethereum on github stars in terms of developer popularity, the fastest growing open source software ever with the biggest community building. The craziest things. Wow. Um Now, one of the things when stability of 1.0 came out, that was a little bit controversial is that unlike some of your competitors like dolly, you didn't apply any filters to either the training data or the output to make sure that people couldn't generate pornography or other sort of not safe for work images. Now, with the 2.0 version that was just debuted last week, I believe you, you've actually rolled that back a bit. You have applied some filters that make it much harder to generate pornography. Why the reversal on that? So the first part was actually a research initiative led by the LMU team at the University of Munich with collaboration from ourselves runway ML and some others that providing the main compute infrastructure. And Robin Rombach, our lead GS of code are doing it and they said, let's start it as a base and make it just an image of the internet and see what we can do and that's what came out. But we put a safety filter on it and an ethical use license as well to limit the amount of output for our own release of stability. We decided and listened to the feedback that we would filter out, not safe for work elements so more people could use it. We decided that we would also make it a more open dataset. So we could really understand how the model worked and we millions of dollars kind of doing that as well, which is now results in the model that anyone can use in a commercial and corporate setting. Great. Well, let's let's, let's see the model and I this is a special demonstration. This is the first time I think you've shown this particular version of the model. Can you, can you explain what you're going to be showing us? Sure. So basically, um we released 2.0 12 days ago, we released the original 1 August 23rd, released 2.1 tomorrow. Um This is the pace of kind of innovation here. 2.1 is very special because when we released 1.01 0.4 actually in August 23rd, it was 5.6 seconds for a generation. We release that release that to 0.9 seconds for an image. Um just a few days ago as of next week, it will be 30 images in a second. So we designed it to be fast and now we've got to real time. This unfortunately isn't quite real time. Um But we've got the one second generation and again, it can generate anything. So you take words, you put it in and for example, is on stage screen. Yes, you can do a lovely Labrador, right? Labradoodle. We like doodles, Labradoodle. And then in a second, it generates a picture of multiple lovely labradoodles with a hat and then boom, it generates it with a hat. I think it's just loading on the internet and then we can do by Van Gogh and it becomes by Van Gogh. Um So this is actually a very interesting thing because 100,000 gigabytes of images compressed into a 1.6 gigabyte file that works on your Macbook. And as of last week, it works natively. So you'll be able to generate these images in a second in a stained glass window. So you can generate just about anything because it understands 2 billion different concepts. And this is just unimaginable, right? And like I said, as of next week, it will be 30 frames. A 2nd 2.1 adds back a lot of the artistic elements. We got some lovely stained glass windows. Isn't that nice? But then it's also very flexible. So for example, what we can do, oops the wrong way we can do. Let's do nine images. Here. We took four images of Jeremy and we decided to create a Jeremy Khan model. This is just with four images. Who knows if this will work? Uh a picture of Jeremy Khan as an astronaut. You want to go to space and then we can do HD map painting. Why not live demo of a new technology who sees knows if it will work or not. We'll find out in a second. The suspense is killing me. OK. OK. How are you doing there? Maybe we should make it professional. Actually, it wasn't professional enough. So it's interesting because it understands all these kind of modifiers. If you do a professional, not professional, you also do negative prompts, you can say not ugly and it becomes not ugly. Um You gotta do not ugly for me. Definitely Jeremy like any of these like, how's this one here? Yeah, not bad. It's fine. Not bad. There we go. Just the four images. We've got Jeremy there and um now we obviously do by banker. There we go, right? So I think, you know, super powerful technology, they can create any image unimaginable just a year ago. That that's not me. That's not you. Another Jeremy's emerged. So mat professional, let's do photo with some 35 millimeter studio lighting to make it good. So, I mean, I think this is the thing, this technology is still nascent, right? This is unimaginable. Just a few months ago, let alone in the last year and it's just evolving at such a rapid rate. Still not, well, it's closer, but this technology is designed to be kind of zero shot where you just put something in, you get out, it's part of a whole design process. So we've seen people take this and create entire movies from it because they take frame by frame like into the Spider Verse and they transform it into an anime movie. We've seen people take this and create architectural designs at scale and you know, again, it's part of the process. It's this amazingly powerful tiny file, we put it out to the world and now people are developing around it and developing an ecosystem that they can then take and build really useful applications because that's what you need actual useful things. This is why the number one App Store app one app Store of the year is Wombo, which uses our technology. Number one on the App Store right now is Lenzo which creates these types of customized profiles based on this version of the technology as well. So I think it's going to be very interesting to see how this develops going forward. Absolutely. And there, I mean, there are skeptics out here sort of would say this is great. It's a lot of pretty pictures. But what's, you know, what's the real business impact of this? So like if you look at movies for example, this will cut down movie generation time by half. So we have an example. Our creative director is shooting a movie in a desert about serial killer with some very famous acts and actresses. Serial killer had a book of all the victims that he had and they were gonna have to hire 25 actresses to fill that book, put them in seventies makeup and you know, fly them out sag rates, make up all this kind of thing. $600,000. He trained a custom model overnight on seventies pictures, it was done. Save $600,000. Wow, that's just one example when you look at video games, movies, other things that transforms in any type of design process, we working with automotive manufacturers to create brand new car designs because you can do these kind of dynamically, you know, you can just type in like an awesome super car in San Francisco and it will generate those but you can customize it with your own assets. This thing when you take the standard model and customize it, it becomes super duper powerful. That's an awesome supercar in San Francisco. Yeah. Wow. And are you, are you thinking of using this beyond just images? Are you thinking about using the same underlying technology for generating other sorts of things? Yeah. So we have the most popular language model in the world G TJ Neo and others from our Luther A I kind of community and we actually do models of every single type. So we have audio models that can generate any type of music. We've got 3D models coming video models. Ultimately, the whole thing is that there's no barrier to creation when you have this next generation of models. Because we move from a big data, we took big data extrapolated for very specific models to these very general models that understand principles like understands the word Supercar, San Francisco, we Twitter about Jeremy, maybe it needs more than four images, but we have bigger Jeremy, right? And this is the next stage in A I and the next stage in real usability of this technology, right? And for you guys, it stability, what's your kind of business model going to be going forward every single creative industry in the world is going to have to react to this because like I said, as of next week, this is 30 images. A second which is basically video brand new experience where you can create anything dynamically. We're going to go to them, take the extra bytes, petabytes of data and convert them into models because again, 100,000 gigabytes to 1.6 gigabytes. Something quite crazy. Yeah, absolutely. Um I said I wasn't going to take questions from the audience, but I think I will because we have a little bit of time who has questions for Ahmad? Anyone have question right down here in front, wait for the mic to get to you, please and please stand and say who you are. What would you say to you? Can you, can you tell us who you are? Sorry? My name is Ian Util Ceo of attention live. What would you say to creatives? Because this can either augment a creatives development of what they do or they can kind of feel like, oh no, this is gonna take my job. So on the optimistic perspective, how are you encouraging creative and creative teams to leverage this technology to do what they do better? So I think this is the thing, what is creativity, right? So this can mash together concepts that so we still have an iterative process to get there. There is just one part of the process. So the ideation phase, the fine tuning phase, the kind of like we have text to image, image, to text in painting so I can remove, I could say it make that car blue and it will make that car blue. In the next version, you have to forget how to integrate this into your processes just as you did with after effects Photoshop and all these other tools because it's just a tool, it doesn't do anything by itself. So what we found is the best artists got even better. And this is what we saw where there was an A I that uh beat Lisa Dollar Go, who was the best Go player? The average Go score. It looks like this over time. Is that because they all got better because they use the technology. This is how you view it as a tool to extend and embrace it because it's here and you should use it right now and then you'll outperform all of your competitors. All other questions over here. Hi there, Ahmad Rishi Gar, a partner at the Mayfield Fund. Um uh You've got a lot of other projects on the website that you're working on. You've obviously started with image generation. But one of the things that's captivated everyone's mind is text to text threading and text generation because of the broad applications. Could you talk a little bit about the future horizon for stability and all the different applications you want to build into and how and how kind of starting with image get you to a place where you can do multimodal applications over time. So that's as a business perspective. When I saw image, I saw it as we going from a 10% to a 90%. Like if I've shown you this a year ago, it's impossible, right? And the fact that it's real time now is also impossible. So you wake up and do some impossible things before breakfast. Text. By contrast, we got 85% to 95% there. And I struggle to see where the edge is going to be when you have a chat bot that can do anything like you use chat GPT somewhere over now. It's fantastic. All the A I models will be like that. Whereas an image I saw that there are giant asset stores of IP that need to be converted because you'll always want Mickey Mouse or Halo or any of these other things. So when you can create everything, something becomes valuable, but it's not enough just to have image because image will be standardized. Instead, what we're doing is we're building the ready player. One oasis hock where anyone can create anything they can imagine. And that's not only media, no more powerpoints, no more slide creation in a few years, just describe it and say make it more peppy, make it nicer and it'll do it automatically recreate Game of Thrones season eight because it was a bit rubbish. You know, these types of things is what will enable seamlessly. And can you talk a little bit about how you build stability itself? Because I think a lot of people were surprised you guys almost came out of nowhere with this fantastic technology. I think there are a lot of businesses out there where they, you know, they talk all the time about the struggle they have hiring A I talent. You've obviously been able to build, you know, really cutting edge stuff. How did you do it? So we built a community when some of the large open source models were open source, the big entities could do it. And I looked at where's the arbitrage ties and the gaps. The gap was super compute data and talent. The talent came from our communities and we hired primo from our communities. So the 1st 96 employees, only one phd so differentiated talent then data, we built large open source data sets like lion 5.6 billion images. The previous launch was 100 million to power this. And then finally super compute worked closely with Amazon to build the 10th fastest public supercomputer in the world. And so that was overwhelming amounts relative to what happened. So 4100 the fastest in the UK was 640. And that unlocked all of this because we allowed people to experiment and play and figure out and make mistakes that then optimized for this crazy model right here. But also other models that were released in our generation to the new language ones, to the others because you need all the components all in one. And that's why creative stability to be like this foundation layer for next generation generative A I and you now have a sort of a partnership with Aws. Is that right? Yeah, so Aws to be a long term partner, now it's accelerating up, we're going to power Generative A I through kind of Amazon and make this available to everyone. It becomes a differentiator for different people's clouds. Obviously, Azure has got their relationship with open A I AWS is you guys, it's kind of an interesting thing. Exactly. And you know, and there, there's different approaches like Azure is quite integrated. Open air is very close to Microsoft. Whereas we're just trying to get this out to the entire world. Great Ahmad, thank you so much for joining us. This has been fantastic, really exciting stuff.

Try this catch-all exercise to boost your stability, combat lower back pain, and improve your running and walking. Austin Afshar from Spear Physical Therapy explains.

Improve your stability with this one minute exercise

Health

Follow these simple steps to improve your stability and overall fitness.

This exercise is a catch all whether you have low back pain balance issues or you want to improve your walking or running. Try this to start. You want to stand up nice and tall if you can grab something in front of you do. So especially if you have balance issues. If not, you don't need to, you want to think about a clock in front of you. 12 o'clock is in front, three o'clock is by my right leg. Nine o'clock is by my left leg. If I want to work on my right leg, I'm going to kick my leg out to the side on a diagonal, aiming for between three and four o'clock when focusing on your left leg. Think about aiming between eight and nine o'clock, three sets of 10 is ideal for this exercise throughout the day. Again, this exercise is great for anyone who is experiencing low back pain. If you have balance issues or you want to improve walking or running.

Stability AI

Stability AI in Fortune Rankings

Videos