Imagine a future with unlimited bandwidth, storage and compute. Now, what is possible?
Today in health, it a world in which data begets data. Can't wait to talk about this. My name is bill Russell. I'm a former CIO for a 16 hospital system and creator this week health, a set of channels dedicated to keeping health it staff current. And engaged. We want to thank our show sponsors who are invested in developing the next generation of health leaders.
Short test artist, site, parlance, and service. Now check them out at this week. health.com/today. Having a child with cancer is one of the most painful and difficult situations a family could face in 2023 to celebrate five years at this week health we are working to give back. We are partnering with Alex's lemonade. Stand all year long.
We have a goal to raise $50,000 from our community. And we are so excited to be up over $40,000 for the year. So far, we asked you to join us, hit our website in the top right hand column. You're going to see a logo for the lemonade stand. Go ahead and click on that to give today. We believe in the generosity of our community. And we thank you in advance.
All right. I came across this article and I really, really liked it. And it got me thinking. , the title of the article is data and compute are the ultimate flywheel. And it's on. It's a blog post. , every dot Tio. , and anyway, just search for data and compute or the ultimate flywheel. And, ,
He talks about what's what's valuable and why in an AI driven world. And this person talks about the data explosion that's happened. He talks historically about some of the things that's happened in his life. The, , but he starts with this in the 2010s. This was the decade of the data explosion. The world began to create log and use more data than ever before. And the big winners in the tech industry, whether social media companies like Metta.
Or e-commerce giants like Amazon made the most of it. But the dawning AI era is changing the playing field data and compute have created a flywheel driven by language models that generate more digital information than ever before. This shifts. Where value sits in software ecosystems. I present key opportunities for large incumbents and new startups. I'm going to come back to that phrase because it's, it's, I think the crux of what we're going to be talking about.
So he talks about, he started an IBM PC with 30. , megabytes of hardness storage. And when you're writing programs at that point, you had very little Ram, you had very little storage, you had to be very conscious of, , memory leaks and all sorts of things like that. Decade later, still have that kind of thing. You know, that was the, what was that? IBM PCs. That was the late eighties.
Then you have the nineties early 2000, we're still worried about, , about data structures and, and those kinds of things. But we had to be really careful. And then he says, , here we are a decade later. And at his new FinTech startup. He said, I didn't care about the data. We saved everything, not just the data assets that form the core of our business with all of their updates and vintages and versions.
But also all of our usage logs and API records and customer reports and website patterns, everything, they saved everything. So what happened? And he asked us. Is this chart. It's a, it's a really good chart. I'm going to have to blow it up here. Let's see. There we go. And it talks about in 1984, the cost of a terabyte of memory would have been a billion dollars, a billion dollars.
Kid you not. , flash Ram doesn't come out to 2005. Memory goes all the way back to there. So flat. So memory goes from a billion dollars per terabyte in 1984 to 2012. Memory is. Yeah. I don't know, much less. Well, 2012, it's much less. It's like a somewhere between a thousand and $10,000. , per terabyte.
Disk storage goes from about a a hundred million per terabyte in 1984 down to well below. , gosh, if I had to figure out what that is, it's less than $50 per terabyte in 2012. And all of these numbers continue to go down. This is, and I've shared this before on the, , on the show. I remember reading the road ahead by bill gates. And one of the things he said is in order to predict the future, all you have to do is, is continue to take all of these things on there.
Logical conclusion, which is the cost of storage is going down. The amount of storage is going up and the ability of miniaturization is going down and the compute is doing nothing but going up and the cost of that computer is going down. So if you had unlimited compute, unlimited storage, , unlimited processing power. That's compute.
, unlimited bandwidth. If you had all those things down unlimited. , fashion. Then you can make predictions. And that's what the road ahead was about, was about bill gates, talking about the various things that are going to be possible in the future. And he describes essentially what we're living with today.
The iPhone instant access to, , just, , , mass amounts of data and information. So he talks about the fact that the companies that had access to large amounts of data and learned how to utilize it. And to mine that data and not only the, the core data for the business, but also the metadata around that data.
, they were the big winners in the economy. And he talks about advertising at length. I'm going to, I'm going to skip over some of this stuff because you can read the article and then it goes, , these flywheels require up-to-date knowledge of customer locations, purchase and travel habits, store inventories, and route, bruh.
G geographies driver and Carvell ability and a whole lot more. Again, none of this is possible without cheap and plentiful data. And I would argue as I'm reading that in healthcare. Do we have this data? We are so focused on what little data we have about health that I think we, we we're, we're missing the forest because we're looking at the trees. We are missing the fact that in order to impact health, we need so much more data than we currently have.
He goes on. Incidentally, these flywheels are not just powered by data. They generate new data in turn. The data explosion is not just an explosion. It is a genuine chain reaction. Data begets data and that's the world we live in and that data. That we run the analysis on and it creates new data. That becomes data that we should store, and we should mine as well.
So he goes on all these flywheels database business models are essentially software that's no surprise data and software are two sides of the same coin. The software used to optimize businesses is useless without data to apply itself to, and data is worthless without software to interpret and act on more. Generally tools are useless without materials, materials don't have value unless worked on with tools.
Economists call these perfectly complimentary inputs. You need both to generate the desired output and you can't substitute one for the, for the other. And immediate consequence is that if the price of one input falls by a lot, Perhaps due to a positive productivity shock. Then the price of the other is almost certain to rise. Imagine you're a tailor.
To so close you need needles and fabric. And you're limited only by the quantity of each of those. You can afford needles and fabric or complimentary inputs. The process of sewing. Now imagine that the price of the needles plummets due to the new needle, manufacturing technology, your response, use the savings to buy more fabric and thus, so more clothes.
But if every tailor does this. Then the price of fabric will rise. Tailors and consumers are both better off. And the total quantity of clothing produce goes up, but the relative value of the needles and the fabric has changed. For the last decade. Plus the quantity of data in the world has been exploding and it's price.
Therefore implicitly declining.
So based on that analogy, he goes on to make this case. For the last decade. Plus the quantity of data in the world has been exploding. And it's price. Therefore implicitly declining. Software has been relatively scarcity input and its price has increased. You can see this in everything from the salaries of software engineers, to the market cap of the top software companies.
Software ate the world with a huge assessed from cheap plentiful data. And then came GPT and everything changed. Okay. So we built this case that. , all this stuff, we have all this data. This is what just happened at any says. Now we live in a new world. Let me, let me talk a little bit about GPT here. So he goes on to say,
GPT is a child of the data explosion, the flood of new data generated not just by users, but also by content farms and click factories and link bots and overzealous SEO agencies necessitated the invention of new techniques to handle it. A team of researchers at Google wrote attention is all you need the paper that introduced the transformers, architecture underlying pretty much every modern generative AI model.
Although large language models were invented to manage the data spun off by these content worse. They're going to be used for a lot more than just search GPT is a prima fascia. A massive productivity boost for software technologists. Talk about a 10 X programmer. The genius can write high quality code 10 times faster than anybody else.
But thanks to GPT. Every programmer has the potential to be 10 X more productive. Than the baseline from just two years ago. We were about to see the effects, move over data explosion, say hello to compute explosion. The first and perhaps most obvious consequence of the compute revolution is that data just got a whole lot more valuable.
This naturally benefits companies that already own the data. But what's valuable in the AI world is subtly different from what is valuable in the past. Some companies with unique data assets will be able to monetize these assets more effectively. Bloomberg GPT is my favorite example. It's trained on decades of high quality financial data. That few others have.
To quote a regrettably, but understandably anonymous, senior executive. In the fin data industry Bloomberg just bought themselves a 20 year lease of life with this. Other companies will realize that they are sitting on latent data assets, data whose value was unrecognized or at any rate on monetized.
Not any more Reddit for one is a treasure trove of high quality human generated content surface by hugely effective modernization of voting system. But now you have to pay for it.
You don't need huge content archives or expensive training to get me full results. He goes on and talks about some new techniques that are in place. Quantity. Has a quality. That's all its own. But when it comes to training data, that converse is also true data, quality scales, better than data size.
Above a certain Corpus size, the ROI from improving quality, almost always outweighs that from increasing coverage. This suggests that golden data, data of exceptional quality for a given use case is well golden. , he goes on and he talks a lot about this. Eh, a lot about the consequences. Let me give you a couple of these. I'll try to keep it brief. The very best data assets reshaped for AI use cases are the new gold mines, but there are terrific opportunities for picks and shovels specifically designed.
Around increased salience of data and the AI first-world, these tools will be able to build new data assets, connect to existing, so forth and so on. More generally the entire data stack needs to be refactored such that generative models become first-class consumers. As well as producers of data, I think about that.
Your generative AI models are the consumers of data and the generators of new data, dozens of companies are emerging to do precisely this from low level. And for providers. And he names a couple. , to high level content engines. Apart from tooling. There's an entire. , commercial ecosystem waiting to be built around data.
In the age of AI pricing and usage models, compliance and data rights. New generation of data marketplace. Everything needs to be updated. No more content without consent. He goes on to talk about the next thing about the flywheel. The second major consequence of AI is that the quantity of both data and compute in this world is going to increase dramatically.
There's both flywheel acceleration data feeds the compute explosion and the compute feeds and the data explosion. And a direct effect after all generative models, don't just consume data. They produce it right now. The output is mostly ephemeral, but there's already, that's already changing as evermore businesses.
Processes begin to incorporate generative components. And this is where I think we need to start focusing is where can generative AI actually look. At our process data and suggest improvements. And I think there's going to be models that come out that look at healthcare and look at the overall efficiency of healthcare and then suggest.
, changes. And then he goes on to talk about the confidence change, compute all things, new abundance, a new scarcity. So, , there's a lot in this article and actually it goes on even from there. So I highly recommend this article. Let me give you a little bit of my takeaways. Oh, by the way, the article is data and compute are the ultimate flywheel.
And, , , I come back to this first race, but the daunting AI era is changing the playing field data and compute have created a flywheel. Driven by language models that generates more digital information than ever before. This shifts where value sits in software ecosystems and presents key opportunities for large incumbents and new startups.
We need to keep our eye on where this is going generative. AI models will be the primary consumer of data in our health systems. They will also be the creator of data in our health systems. Even the ugly data that we've had with some of the use cases we've heard people talk about at this point are taking the entire medical record, pointing the gender of am. I will add it, it consumes the whole thing and then creates a.
An understandable. , summary of that record. Or pulls out the salient points for the clinician of that record, because it knows how to identify the salient points. So instead of pretending that the doctor has read the umpteen million PDFs that are in there, we would actually have something, the computer model read those PDFs that are in there and surface the relevant information.
So it will be a consumer of information, but it will also be a producer of information. Those records will then be used in many different ways. There'll be used to educate the patient P available to the patient for the patient to share with another clinician. It could be the, the, , The title page, if you will, for the book or the jacket cover for your health novel, if you will, and they can read the jacket cover, and then they can click into the various things and go to, I don't know, with Tesco, maybe.
, A Q Hannon. And capture that data that they need. And pull that into the, or their health system. , so I keep coming back to this. Generative AI is going to change everything. It changes everything today. And we should keep our eyes focused on this because everybody I'm talking to in every other industry is saying, this is going to change finance. It's going to change marketing. It is going to change retail. And if things are going to change that rapidly in these other spaces, in these other industries,
It is going to rapidly change healthcare now, rapid in healthcare and rapid and retail and FinTech, as we know. Are are completely different timelines. Rapid in healthcare would be, I say a three-year time horizon. In three years, if you are not utilizing generative AI in healthcare, you're going to be falling behind.
You're definitely going to be falling behind the front runners, the Mayos and the Stanfords and the UCS, UC Davis, UC. , San Diego and the rest. Of the organizations that are getting ahead of this. , but then you're going to start falling behind your local competitors and keep in mind that some of your local competitors are also going to be really good at this. It is going to be the.
, one Medical's under Amazon. It's going to be the CVS. It's definitely going to be the Walmarts they're already playing with this. It's something we should keep our eyes on and try to stay ahead of, sorry. I went a little long today. , seem to be doing that. I'll tell you what I'll do two short ones over the next two days.
Do you know, if so that's all for today, by the way, if you know someone that might benefit from our channel, please forward them a note. They can subscribe on our website this week out.com or wherever you listen to podcasts, apple, Google, overcast, Spotify. Stitcher that's as fast as I can say that you get the picture.
We want to thank our channel sponsors who are invested in our mission to develop the next generation of health leaders, short tests, 📍 artist, site, parlance, and service. Now. Check them out at this week. health.com/today. Thanks for listening. That's all for now.