Brainstorm AI 2024: Building A Better Model For Creativity

Fortune

Presenter: Bill Gross, Founder and CEO, ProRata AI; Founder and Chairman, Idealab

Transcript

00:00I've been an entrepreneur all my life, and as a 16-year-old in Pasadena, California,

00:05I was reading Fortune magazine, looking up to all the incredible businesses that they

00:09spoke about.

00:10And then I had a chance to come to all these Fortune conferences and learn so much from

00:13all of you and all the people at Fortune magazine.

00:16And now here I am on stage telling you about a new idea, so it's really thrilling to be

00:18here.

00:19Thank you so much.

00:20What I want to talk about today is the disruptive world we're living in right now, which is

00:25just more disruptive than anything I've ever seen.

00:29Could you cue to my slides, please?

00:31We are living in wildly disruptive times, and the question is, how will creators survive

00:36in these times?

00:37Think back to all the different technological revolutions and how creators had to adapt

00:41to be able to thrive.

00:43Back to the printing press, radio, television, and, of course, the internet.

00:48But I think these times are even different than anything we've ever seen before.

00:52I think this AI disruption will dwarf anything we've seen in our lifetimes.

00:55I put up on this slide four things that I saw in my lifetime, which when I saw them,

01:01I thought the world would never be the same.

01:04I go back to the IBM PC in 1981.

01:06It actually came out the very month I graduated from college.

01:10I went down to my local computer land store in Pasadena, California, bought a $5,000 monochrome

01:15machine with two floppy drives and 256K of RAM for $5,000.

01:20No hard drive because the hard drive hadn't been invented yet.

01:22But it completely blew my mind, and, of course, it changed the world.

01:26It led to a whole new form of creativity, code, and led to trillions of dollars of wealth

01:30creation.

01:31Then, Netscape in 1994.

01:33When I saw that, I also felt the world will never be the same.

01:36In fact, I was running a company called Knowledge Adventure at the time, selling educational

01:40software for kids, and we sold that company in 1995, and now I had all these new ideas

01:46for internet companies that you could build on top of the browser.

01:49I put down my favorite 12 ideas, and I was trying to select which one I wanted to start,

01:53and I actually decided to start all 12.

01:55The way I did that was by starting Idealab, a technology incubator where I could start

01:58all 12 at once under one roof.

02:01One of the companies we started, based on the browser, was a company called GoTo.

02:05We invented pay-per-click advertising, and it completely changed the way monetization

02:09worked on the internet.

02:10We were really proud of that.

02:12But that just shows how new models are needed whenever there's a new platform, a new technology.

02:16Of course, the mobile revolution has been talked about already today, that changed all

02:20of our lives.

02:21In 1994, when I started looking at Netscape and seeing how powerful it was, there were

02:2630 million browsers in the world.

02:27Think about how pitiful small that is compared to the number of mobile devices that are on

02:32the planet now, 5 billion supercomputers in everybody's pockets all over the world.

02:36So all these changes were so dramatic, and then November 2022, we're at the two-year

02:41anniversary of ChatGPT being re-released.

02:44I remember playing it with that one day and saying, this is going to change everything,

02:49and of course it has.

02:50It's led to trillion dollars of wealth creation and expenditures in just two years, not like

02:55the PC, which took two decades for that kind of impact to be felt.

02:59So when you have something that big, why does it cost so much change?

03:03Well hugely disruptive changes occur when the cost of something approaches zero.

03:08With the internet, the cost of distribution approached zero.

03:11With the cloud, the cost of storage approached zero.

03:13And with generative AI, the cost of knowledge or knowledge creation is approaching zero.

03:17So what does that mean for content creators when the costs change that dramatically?

03:21This disruptive moment is really, really large.

03:23Well, until now, one side of the equation, some of the generative AI companies are saying

03:29the content must be free.

03:31The other side of the equation, the content producers are suing.

03:34On the one side, you have Mustafa from Microsoft saying, everything on the internet can be

03:38used for free to train AI models.

03:40Or Sam who's saying, we need this copyrighted material for free to make this business model

03:44work.

03:45On the other side, you have the New York Times, Dow Jones, and many, many other companies

03:48suing a lot of the generative AI companies because they feel their content is being stolen.

03:53In a funny tweet, Ariel Dumas says, I've been told to stop stealing muffins from the bakery.

03:59Unfortunately, it's the only way to keep my lucrative muffin stand in business.

04:02Everyone is fine with this.

04:04And of course, there's sarcasm here, but it's maybe not that far from the truth about companies

04:08lifting other people's content and using it without paying for it.

04:11And that leads to a dilemma today.

04:13Well, in almost all other forms of digitization, revenue is shared.

04:19I'll give you some examples right here.

04:21On the App Store and on Google Play, revenue shared with creators.

04:25On Spotify, revenue shared with the artists.

04:28In fact, Spotify shows that you can make a business like this work even with significant

04:33revenue share.

04:34Spotify's market cap just today crossed $100 billion.

04:37They had $16 billion of revenue, and $10 billion of revenue was shared with creators.

04:42It was shared based on counting the streams that were played.

04:45YouTube shares revenue, full 50-50 revenue share, sharing revenue with the person who

04:50created the stream.

04:51Apple Music and also Apple News.

04:53Apple News is another example of another method of counting.

04:57You can count by streams, you can count by views or downloads.

05:00Apple News counts by minutes, minutes spent in different media content.

05:04And Apple News took in $2.3 billion of subscriptions last year, and they paid out $1.15 billion

05:09to all the content partners who provided the content.

05:12And even in the physical world, for offline distribution, revenue is shared.

05:16Companies like Nielsen, ASCAP, BMI count and share revenue in those use cases as well.

05:24So why are there no revenue shared standard in generative AI?

05:29It's because there's no simple streams or views or downloads to count.

05:32The answers that come from generative AI are an amalgam of different content.

05:38But what if you can reverse engineer the output of generative AI and figure out what content

05:42contributed to the answer?

05:44That would enable a fair revenue share.

05:46So right now, some of the AI companies are saying, we don't want to pay because we can't

05:49afford to, or we don't think we have to.

05:51But some generative AI companies are saying, we can't pay because we don't know how to.

05:54Well, what if there are a way to solve how to?

05:56Well, that's what we set out to do.

05:58We set out to create an algorithm that could take a look at the output of generative AI,

06:02whether it's text, images, movies, or music, analyze it using AI, and figure out where

06:07that content came from.

06:08And not only figure out where it came from, figure out the percentage that came from each

06:11place.

06:13And just today, we launched this new search product called GIST.

06:17At gist.ai, you can go, type in a query, and you see an attribution bar at the top.

06:23That attribution bar at the very top is breaking down the sources and how they contributed

06:28to the answer.

06:29So, for example, in this particular question, 34% came from the Atlantic, 25% came from

06:35Time, 25% came from Fortune, Fortune's one of our partners, and 16% came from The Guardian.

06:40And all of the sources are called out clearly on the right.

06:44This then allows us to share revenue 50-50 with all those partners on that exact pro-rata

06:49basis.

06:50We think this is fair.

06:51We think this is a great way to encourage monetization of content in the generative

06:55AI era, but also encourage, by providing a business model, for more creators to continue

07:01to make great content to power generative AI.

07:04If generative AI only starts using synthetic data for its training, the result in quality

07:08will go way down.

07:10But if content creators have the incentive to keep on creating great content, the quality

07:14will go up.

07:15So, we really believe this is the future, and we really want to try and make something

07:18like this be the standard.

07:20We really feel that generative AI needs to be ethical and fair to make this grow and

07:24thrive.

07:25I showed you the example with text.

07:28It also could be done for images.

07:29Here's an example.

07:30The prompt that was typed in was, generate an image of a mass red superhero that can

07:34climb walls.

07:36Meta generated this image.

07:37We look at that image, we analyze it, we look at all the source images, and we figured out

07:41that this came 90.3% from Marvel and 6.2% from DC Entertainment, again, giving us the

07:47exact percentages so the revenue could be shared.

07:51So, in generative AI, it is now fully possible to look at the output and figure out who contributed

07:56uniquely to that content.

07:59So far, hundreds of publications, authors, and musicians, and artists have joined our

08:02system.

08:03Actually, 400 as of today.

08:0420 more just joined this morning.

08:06We made an announcement this morning of new ones, including a company called Lee Enterprises,

08:10which has 71 markets around the United States, and we're working with them to revolutionize

08:15the way local news is surfaced in generative AI, and also the way local news can monetize

08:21with generative advertising as well.

08:23We really feel that this is a new frontier.

08:25If you think back to search and how it's gone through multiple iterations, search 1.0 was

08:31monetized with banner ads.

08:33Think back to Excite, and Lycos, and Alta Vista, and Yahoo.

08:37Search 2.0 was monetized with pay-per-click.

08:39I believe that search 3.0, which we're seeing right now, will just deliver you answers directly.

08:44You won't have to click, but it'll be monetized with pay-per-use.

08:48The pay-per-use is the ability for the content creators to make some revenue every time their

08:53content is used.

08:54Of course, their content also can be used for training, but that's a one-time fee just

08:58for training the model.

08:59Pay-per-use enables the content creators to have a sustainable revenue model that can

09:04be ongoing.

09:06So ProRata.ai enables ProRata compensation on this pay-per-use basis, and I really feel

09:11I'm so passionate about this because I really want to make a world where content creators

09:15can thrive, where it's fair and ethical for everybody, and everybody can participate.

09:20Wherever the content is created anywhere on earth, it can be surfaced in answers to help

09:24you have the best content possible and to make a revenue stream monetizable.

09:29I'd love to work with any of you here.

09:31I'd love to have some of you join our ProRata platform, but mostly I would like feedback

09:34and ideas on how to make this a more successful, thriving ecosystem around generative AI.

09:40Feel free to reach out to me directly on Bill at ProRata.ai.

09:43Thank you very much.

09:44You've been a great audience.

Category

Transcript

Recommended