Model Transparency and AI Browser War
Key Points
- The host argues that true model transparency requires publicly releasing the training data and model weights, not just using closed‑source models.
- The episode of “Mixture of Experts” brings together experts (Chris Hay, Kate Soule, Aaron Baughman) to discuss AI topics such as transparency, AI scrapers, and emerging technologies.
- A light‑hearted “browser war” segment highlights each guest’s preferred web browser and recalls historic battles between Netscape, Internet Explorer, Firefox, and Chrome.
- The conversation turns to a new wave of AI‑driven browsers—like Perplexity’s Comet and rumored OpenAI offerings—emphasizing that modern browser competition is a key revenue driver for tech companies.
Sections
- Call for Model Transparency - A host argues that AI models must be open—publishing training data and weights for public evaluation—and previews a panel discussion on transparency, AI scraping, and related topics.
- Search Friction and Platform Control - The speaker describes how AI‑powered sites like Perplexity and ChatGPT want users to access their answers directly in the browser, reducing navigation steps, and discusses the resulting battles over user control, ad revenue, and Google’s likely reaction.
- AI Integration Threatens Browser Privacy - The speaker requests Commet browser access and then discusses how the rise of AI services within browsers could undermine the privacy safeguards traditionally promoted by browsers such as Brave.
- AI Subscription Inequality Debate - Participants argue that AI services will retain paid tiers, creating a two‑class system where wealthier users access premium models while others receive limited, ad‑supported experiences.
- AI Browser Competition and Transparency Push - The speakers discuss the emerging rivalry among AI‑integrated browsers shaping online knowledge access, then shift to Anthropic’s proposal for stringent transparency regulations targeting large frontier‑model developers.
- Balancing AI Startup Regulation - The speaker debates whether emerging AI startups should face strict safety and transparency rules akin to established providers, arguing for standards even for small companies, especially in high‑risk fields like medical diagnostics.
- Debating Transparency vs Safety - The speakers argue that while greater openness could aid problem‑solving, it may also spread dangerous AI capabilities and create unfair competitive advantages.
- Balancing Transparency with Commercial Constraints - The speaker argues that while openness is valuable for responsible AI, disclosure should be guided by downstream risk and commercial realities, favoring selective transparency over blanket release of model weights.
- Shared Responsibility for AI Use - The speaker stresses that both AI developers and users must adhere to terms and guardrails, cautions against overly permissive transparency exemptions for smaller firms, and asks how Granite approaches model‑card transparency.
- Granite Bug Bounty & AI Scraper Alert - The speaker announces a forthcoming white‑hat hacking and bug‑bounty program for Granite, encourages community reporting, and then highlights Cloudflare’s emerging issue with AI‑driven web scrapers generating high traffic and load.
- Balancing Content Control and AI Access - The speaker discusses the tension between Cloudflare’s permission‑based internet model that empowers creators and the need for open data to train AI, proposing decentralized tools to maintain control while warning against a tragedy of the commons.
- Balancing Data Access and Fair Compensation - The speakers discuss how the AI industry's reliance on web‑scraped data is prompting creators to restrict access, urging a healthier dialogue among content providers, data aggregators like Common Crawl, and model developers to ensure more equitable distribution of the value derived from open internet resources.
- Evolving Content Formats and Monetization - The speaker discusses how emerging markup languages like markdown could let bots handle new, payment‑enabled content protocols, applauds Cloudflare’s move to safeguard creators’ revenue, and critiques the current “free‑for‑all” web environment that undermines monetization.
- Real‑Time Wimbledon Fan Experience - The team highlights an AI‑powered, large‑scale system that offers live match chat and real‑time win‑probability estimates to enhance Wimbledon spectators’ experience.
- Predictive Modeling for Fan Experience - The speaker explains how a pre‑match win‑probability model using traditional machine learning (SVMs, decision trees) underpins a reimagined, data‑driven spectator experience, and probes fan adoption of these tools while citing an open‑source ACM KDD paper on Barcelona.
- Predictive Models and Sports Thrill - The speakers discuss how increasingly accurate predictions could diminish the excitement of sports by pre‑determining outcomes, while also creating new hype, gambling implications, educational tools, and even influencing player behavior.
- Cricket Stats, Birthmonths, and Trends - A speaker reflects on how birth month influences cricket performance, highlights Sachin Tendulkar as an outlier, and argues that data analytics will reshape fan engagement and the future of sports.
Full Transcript
# Model Transparency and AI Browser War **Source:** [https://www.youtube.com/watch?v=ieR3u8sUX-U](https://www.youtube.com/watch?v=ieR3u8sUX-U) **Duration:** 00:51:48 ## Summary - The host argues that true model transparency requires publicly releasing the training data and model weights, not just using closed‑source models. - The episode of “Mixture of Experts” brings together experts (Chris Hay, Kate Soule, Aaron Baughman) to discuss AI topics such as transparency, AI scrapers, and emerging technologies. - A light‑hearted “browser war” segment highlights each guest’s preferred web browser and recalls historic battles between Netscape, Internet Explorer, Firefox, and Chrome. - The conversation turns to a new wave of AI‑driven browsers—like Perplexity’s Comet and rumored OpenAI offerings—emphasizing that modern browser competition is a key revenue driver for tech companies. ## Sections - [00:00:00](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=0s) **Call for Model Transparency** - A host argues that AI models must be open—publishing training data and weights for public evaluation—and previews a panel discussion on transparency, AI scraping, and related topics. - [00:03:02](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=182s) **Search Friction and Platform Control** - The speaker describes how AI‑powered sites like Perplexity and ChatGPT want users to access their answers directly in the browser, reducing navigation steps, and discusses the resulting battles over user control, ad revenue, and Google’s likely reaction. - [00:06:22](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=382s) **AI Integration Threatens Browser Privacy** - The speaker requests Commet browser access and then discusses how the rise of AI services within browsers could undermine the privacy safeguards traditionally promoted by browsers such as Brave. - [00:09:42](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=582s) **AI Subscription Inequality Debate** - Participants argue that AI services will retain paid tiers, creating a two‑class system where wealthier users access premium models while others receive limited, ad‑supported experiences. - [00:12:45](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=765s) **AI Browser Competition and Transparency Push** - The speakers discuss the emerging rivalry among AI‑integrated browsers shaping online knowledge access, then shift to Anthropic’s proposal for stringent transparency regulations targeting large frontier‑model developers. - [00:16:03](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=963s) **Balancing AI Startup Regulation** - The speaker debates whether emerging AI startups should face strict safety and transparency rules akin to established providers, arguing for standards even for small companies, especially in high‑risk fields like medical diagnostics. - [00:19:14](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=1154s) **Debating Transparency vs Safety** - The speakers argue that while greater openness could aid problem‑solving, it may also spread dangerous AI capabilities and create unfair competitive advantages. - [00:22:21](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=1341s) **Balancing Transparency with Commercial Constraints** - The speaker argues that while openness is valuable for responsible AI, disclosure should be guided by downstream risk and commercial realities, favoring selective transparency over blanket release of model weights. - [00:25:23](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=1523s) **Shared Responsibility for AI Use** - The speaker stresses that both AI developers and users must adhere to terms and guardrails, cautions against overly permissive transparency exemptions for smaller firms, and asks how Granite approaches model‑card transparency. - [00:28:24](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=1704s) **Granite Bug Bounty & AI Scraper Alert** - The speaker announces a forthcoming white‑hat hacking and bug‑bounty program for Granite, encourages community reporting, and then highlights Cloudflare’s emerging issue with AI‑driven web scrapers generating high traffic and load. - [00:31:30](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=1890s) **Balancing Content Control and AI Access** - The speaker discusses the tension between Cloudflare’s permission‑based internet model that empowers creators and the need for open data to train AI, proposing decentralized tools to maintain control while warning against a tragedy of the commons. - [00:34:43](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=2083s) **Balancing Data Access and Fair Compensation** - The speakers discuss how the AI industry's reliance on web‑scraped data is prompting creators to restrict access, urging a healthier dialogue among content providers, data aggregators like Common Crawl, and model developers to ensure more equitable distribution of the value derived from open internet resources. - [00:38:01](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=2281s) **Evolving Content Formats and Monetization** - The speaker discusses how emerging markup languages like markdown could let bots handle new, payment‑enabled content protocols, applauds Cloudflare’s move to safeguard creators’ revenue, and critiques the current “free‑for‑all” web environment that undermines monetization. - [00:41:15](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=2475s) **Real‑Time Wimbledon Fan Experience** - The team highlights an AI‑powered, large‑scale system that offers live match chat and real‑time win‑probability estimates to enhance Wimbledon spectators’ experience. - [00:44:19](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=2659s) **Predictive Modeling for Fan Experience** - The speaker explains how a pre‑match win‑probability model using traditional machine learning (SVMs, decision trees) underpins a reimagined, data‑driven spectator experience, and probes fan adoption of these tools while citing an open‑source ACM KDD paper on Barcelona. - [00:47:23](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=2843s) **Predictive Models and Sports Thrill** - The speakers discuss how increasingly accurate predictions could diminish the excitement of sports by pre‑determining outcomes, while also creating new hype, gambling implications, educational tools, and even influencing player behavior. - [00:50:26](https://www.youtube.com/watch?v=ieR3u8sUX-U&t=3026s) **Cricket Stats, Birthmonths, and Trends** - A speaker reflects on how birth month influences cricket performance, highlights Sachin Tendulkar as an outlier, and argues that data analytics will reshape fan engagement and the future of sports. ## Full Transcript
I just fundamentally disagree.
I don't,
I don't believe in closed models. Right.
I use them, obviously.
But I really think the most way, the best way of being transparence
is publish what data was used to train your model and publish your weights.
And we can all have a good look at it.
And then we can tell you
whether it's a good model or a bad model and you can improve it.
But what if it's locked behind the door?
Then how do we know.
All that and more on today's Mixture of Experts.
I'm Tim Hwang and welcome to Mixture of Experts.
Each week, MoE brings together a crack team of the most brilliant
and entertaining researchers, product leaders, and more to distill down and
chart a path through the ever more complex landscape of artificial intelligence.
Today I'm joined by Chris Hay,
Distinguished Engineer and CTO of Customer Transformation.
Kate Soule, Director of Technical Product Management
for Granite, and Aaron Baughman, IBM Fellow and Master Inventor.
We have a packed episode today.
We're going to talk about model transparency, AI scrapers, Wimbledon,
but first I really want to talk about browsers.
And I guess we'll start with our little round the horn question.
A little bit of a personal question if I can ask.
And so let me know if this is like, you know, too invasive.
But Aaron, I'm curious about like what web browser do you use on a daily basis?
Well, if I'm...
In a conversational mood then I'll use, you know, ChatGPT right?
So that's something that that I enjoy.
If I'm in a search kind of mode and I'll use Chrome.
Cool. That's great.
Chris, how about you?
I'm still on Netscape Navigator.
Old school.
Very nice, I respect that. And Kate, what do you think?
I use Chrome on my laptop and I use Brave on my cellphone.
Okay, nice.
Yeah. I'm a big, Brave fan here too
to the use of the phrase "browser war" is kind of funny in the AI context.
I just want to give a little bit of history here.
I was looking at a little bit before the show started,
so "browser war" has been something that's been like used as a term of art
for competition in the technology space for decades now.
But the first browser war was Internet Explorer versus Netscape Navigator
in the late 90s, happened again in the 2000s between
Internet Explorer, Firefox and Chrome.
And I think it's really interesting that we appear to be on the cusp
of an entirely new browser war.
So Perplexity.
The AI, search company has launched a new AI driven browser called Comet.
And, there also are rumors that OpenAI is working
on, one of these sort of AI driven browsers, as well.
And so, Chris, maybe I'll start with you.
You know, I guess maybe the first thing I think would be good for our listeners
to get intuition is like, why are these cutting edge technology companies
working on a kind of software which is really old, right?
Decades old at this point? Like, why is it important?
I mean, it's super important because that's the path to revenue.
I would love to explain it in a different way, but whoever controls
the pane of glass is is controlling the revenue.
So if we think about how we do search
today, you go to Google, you're going to click on your browser bar.
You don't even go to the website anymore, right?
You type in, you search, get your ten blue links and then you click through.
Right. And really,
everybody wants to be the first place where you go for information, right?
And it's great Perplexity is an amazing site.
It comes back with absolutely awesome stuff.
ChatGPT is the same, but actually there is a friction point, right?
So when I do a search, I open up my browser or, you know,
or on my phone or whatever, and I have to go to this, the site
and you know how I'm usually going to the site, I'm googling
Perplexity because I'm too lazy to do anything else.
Right.
So I put it in the browser bar, you know, I put the word perplexity or ChatGPT,
and then eventually I go click, click and I go to the site.
They don't want that experience, what they want is when you open up
the browser, you are going directly to their pane of glass.
And therefore when you do the interactions and when you do the search,
they are coming back with a search result.
So this is this is all about control.
This is all about ultimately it's going to come back down to ads
at some point and revenue.
So obviously Google's not going to be happy about this.
And there is not a lot they're going to be able to do about it as well
because it can't go all, you know, gunslinging or whatever phrase is.
Right in that sense, because they've just had a whole you gotta sell chrome thing.
Right?
So they of suddenly go, yeah, we're going to get rid of you, you know?
So this is going to be interesting.
Interesting. Aaron, I. Guess question for you.
This is a little bit of a, I think, almost change of tactics.
I don't know if you'd agree
from the frontier AI model companies or I guess Perplexity as well.
Right.
Like companies that are primarily kind of AI first.
Because I think there's maybe a view
that originally like why why rebuild the browser, right.
What we're going to do is just simply everybody's going to adopt chatbots.
And so like what used to be on search will just go over to chat bots
and that will be that.
This seems like now the other way around. Right.
Which is like they're saying, okay,
well we also have to launch in a browser form factor.
And I guess, I mean, in response to that, you're
kind of answer to around the horn question.
It seems like right now you're you're kind of doing both.
And I guess the question is whether or not you think over time
they just converge on the same thing, right?
Like there won't be any browsers anymore.
It really will just be chat bots. Everything.
And what they're calling a browser is really just a chatbot.
Yeah.
I mean, you know, the AI first browser wars are here, right?
I mean, it's great to to watch and be a part of it.
You know, instead of this internet search,
we're now moving towards this conversational.
And you can think of it like this context aware discovery, you know, and
I saw this quote where these AI, you know, first browsers,
it's designed for curiosity and built for answers, right.
Instead of, you know, just going and searching
and trying to link all these data pieces together.
And, and then we have to, to really do the work right.
And, the stakes are just really high.
You know, Chris pointed out that that really, you know,
all these different companies, frontier companies and even established companies
now, or they're they're the jousting for this default entry point to the web.
And, you know, the web itself, internet is changing.
You know, there's a lot of, agent tech protocols that are now, coming out,
you know, so it's agent talking to agent providing,
you know, answers through through conversations, you know, and,
and then and then what I like to do is this hybrid, you know,
kind of searching where I have a traditional search browser that's then,
you know, flanked off on the side by maybe, you know, a OpenAI
or even a Perplexity, kind of, you know, search capability there.
And and yes, I'm still waiting to get access to that Commet browser.
So if anyone from Perplexity is listening, take take me off that waitlist.
Right. So I can do it so I can get in there. Right.
And no pressure. You said.
That's right. Yeah, yeah.
Kate, one dimension to this
I think is kind of interesting is particularly in browsers,
it feels like that there has been this kind of long history of thinking
about the browser as sort of like your privacy defender on the internet.
You mentioned Brave. I'm a big, Brave fan, right?
For folks who don't know, it's a project that was kind of, an offshoot of Chrome.
But the explicit purpose is how do we kind of defends use the browser
as a way of kind of shielding your privacy online, I guess.
What question is like,
should we be worried
about kind of the entrance
of these more AI companies into the space because it feels like the norm?
There hasn't been so strongly set right like that.
It feels like part of the expectation with, say, ChatGPT
is that they get all your transcripts. Right?
And so I don't know how you think that's going to evolve, like in some ways.
Do you think part of this is like resetting the norms of,
I think all the good work that's been done around
browsers or will kind of the browser norms around privacy hold over time?
Yeah, I think it's a really good point and I completely agree.
Ads and revenue is the primary driver here,
but I think there's
a really important driver here that OpenAI is pursuing on data collection.
I feel like I say this every time that I'm, on this show, it's
all about more and more ways.
Yeah.
Open AI can collect data, collect interesting, relevant data
and not have restrictions in terms from other browser providers,
limiting what they can track,
and what they can gather and use.
So I completely agree.
I think this is going to, you know, open up new revenue opportunities,
both from a ads perspective but also from like continuing
to improve the model and continuing to try and create differentiation
in the models that are driving the value that's trying to attract the customers.
So creating a nice little virtuous cycle for them.
And I worry tremendously about the privacy risks.
And as you said, I use Brave for the same exact reason because
I'm fairly privacy, conscious above, you know, my online presence.
And, you know, I don't think the incentives
are going to hold to maintain that level of privacy
with these model providers creating their own browser platform.
Yeah. That's right.
I mean, not to mention I mean you know, ads have been
at the kind of center of the discussion so far.
Like I think one of the things that's kind of nice about the existing generation
of like say chatbot tools is that it's like subscription.
So I feel like I've got a little bit more trust in that than in a world
where they are starting to monetize through ads, and that almost
feels like a changes a lot of the deep kind of value prop, right of like,
why these technologies are really cool and exciting in the first place.
Do you think there will be continue
to be a market for a subscription,
or is that kind of like a little bit like me early in the day?
You need to be able to subscribe to a like a search engine very briefly as well.
And so kind of the question is like whether or not like that holds
as the primary business model.
I think we're going to continue to see subscription where there are specific
terms around how your data is used, particularly for enterprise use cases.
Right. Companies are going to want to pay
OpenAI and others to make sure that their sensitive data is,
you know, handled in a very specific way and format that those terms
do not necessarily apply to, you know, the general web browser
that any of us download and then start to use for opening AI.
So I don't think that the subscription is going to go away.
I think there's,
too much emphasis for enterprise use models on data stewardship
where they're going to use monetize that in, you know, subscription type format.
Yeah. That's right.
I mean, Chris, do you think that I mean, one result of what Kate's saying is almost
like increasing like kind of inequality online, right?
Like they'll be like the class of people who can afford to pay like $200 a month
for the extra pro version or, who knows, $1,000 a month, right?
I know is opening.
I was quoting a little while back,
and then they'll be everybody else who gets like the browser for free,
and it'll be kind of a chatbot
bot experience, but they'll monetize and it'll be ads and all that.
All the time. You're breaking my heart at the moment.
You're you're you're telling me my current predicament, right?
The amount of money
that goes out of my account to AI companies at the moment is incredible.
Like rent, utilities and, you know.
And luckily, my wife doesn't watch this podcast, so, she doesn't know this,
but but no, I think,
I, I think there is kind of a two class system at the moment because
unfortunately, it's not enough to have one subscription.
Right?
You want to be able to go with the latest and greatest of all the models.
And and the reality is with the subscription, you get higher limits.
You get, access to the latest models.
So it is very much a two tier system.
But then these models are hugely expensive to run.
Right? So, you got to pay for it somehow.
I would like to see a world
where, compute becomes ubiquitous and therefore,
in its charge, like
kind of electricity in that sense, and is kind of just a metered.
Right. I think that would be a good thing.
I would like to see more of a grid system for compute,
as opposed to sort of plugging into a wall like an electric charger.
So I think that's, that's kind of, that's kind of where I'd like to see.
But I think we're well off of that.
But this is why open source is so, so important
and why having compute on your laptop is so important.
Right.
And and why it's really important to get the smaller models
as quick and as powerful as possible so that we can open that up.
And, and, and it's not relying
on those with the biggest GPU and those with the biggest compute.
I know that's rich
considering the amount of money I spend on AI, but I but but I really,
really, really think that the compute should be ubiquitous in this sense.
Yeah. That's right.
If it is widely available, you would just use more, right?
You'd be subscribing and also using all the open source models.
Ok, Aaron final thought on this is so in the original browser
war, you had Internet Explorer versus Netscape Navigator,
and in the second battle you had Internet Explorer versus Firefox versus Chrome.
And so I guess so far to this race, we've got comment from Perplexity,
OpenAI probably.
Do you have any other predictions on who's
going to jump in next on the on the great AI browser war?
Yeah, I mean, you know, there's there's always Microsoft's Edge you know
and Copilot there's the browser companies you know Dia there's Arc right.
I mean there's there's many many,
you know, companies that are gonna, you know, jump in.
And I mean, the winner of the AI browser wars.
And there may not be a singular winner here.
It might be a conglomeration or, or an aggregation right, of these types
of AI based browsers, but they will define how we access knowledge
and how we do task and even think online, maybe even for the next decade, you know,
so it's it's going to be fascinating, right, to to watch this.
Yeah. We'll keep an eye on it for sure.
I'm gonna move us on to our next topic.
So interesting post came out of, Anthropic, just recently.
And it's entitled "The Need for Transparency and Frontier AI."
And effectively, what anthropic does is that they say, look,
you know, we're a frontier model company.
We think it's important to be responsible in the space.
And so what we're putting forwards is our proposal on what sort of regulation
in effect should look like around frontier model transparency.
And they list a long list of things that should be done.
Right.
So they say, you know, these transparency regimes
should only apply to the largest model developers.
You know, people need to publish system cards.
They need to make their secure development frameworks public.
They need to protect whistleblowers.
They need transparency standards, kind of a whole sort of regulatory stack
that they kind of propose as kind of the framework here.
And I guess, Kate maybe I'll kick it over to you.
I mean, I'll just play cynic for a moment, right?
Which is, I don't know. I'm a big company.
That's one of the most valuable companies in the world.
Why am I calling for regulation?
Doesn't that just raise the costs of doing business? Like, why?
Why is that a thing that, like, Anthropic would be doing?
I mean, I think anthropic has good motives and have always tried to
on the side of a conservative, responsible approach to AI. So.
So that I don't think we could refute.
But I do worry, and I'm all for transparency.
And I let me just start there.
I do worry
the way that Anthropic has framed some of their suggestions, with cut offs
and only frontier model providers and large language model providers
kind of plays into this, you know, story that anthropic is,
perpetuated and, you know, gotten some flack around that
only big model providers can build safe
AI and can only a few people, namely them, can do this responsibly.
And, you know, I think where they start to go
maybe a little astray in these, proposed requirements
is that they really focus on, okay, we're going to make sure
that the big model providers have to play transparent,
but start up smaller places,
you know, models where there isn't a lot of revenue attached to it.
You don't have to worry.
You can get exempt
that way where, you know, allowing everyone to continue to innovate.
And I think that's the wrong approach.
If we talk about why we need transparency, it's to help people understand
and manage risks.
And risks are
not just based off of the model, they're based off of the application.
And so I don't care.
This model was trained by a tiny startup or trained by Anthropic or trained by IBM.
If it's being used for medical decision making, there is a set of risks and,
a framework that we should be transparent about that
everyone should be able to look at, understand, just like they,
you know, inspect the label on any FDA approved drug or whatever the case may be.
So, you know,
I think they are straying a little bit too close to the only big model.
Providers can do this safely, responsibly.
And I really think we should be having a conversation.
How do you have transparency around risks that are application
based, not provider based or model based and that type of thing?
I mean, I guess I.
I'm sympathetic a little bit to like the little guy here I guess.
Right.
Like if you're a startup and you're like, oh man, we want to deploy this
AI technology.
And there's all these regulations around safety and transparency.
It does make it harder to operate.
I guess you're kind of saying like,
maybe you should just, you know, man up and do it.
But like, I'm kind of curious about like, how you think about that because I think
it's almost coming from a place of like, trying to, like, lower the burden.
But I guess maybe you're kind of saying you're a little bit not too sympathetic
if. You're a startup playing and you know, medical diagnostics space
where you're, you know, making recommendations on patient care,
I sorry, I don't care if you're like a little startup.
There should be basic requirements.
We don't give free passes to companies, you know, developing drugs
just because they're new.
They still have to go through the standard
trials process to bring a drug to market.
So but that doesn't mean that there aren't plenty of ways in areas for startups
and other, players that are smaller trying to get their foot to start building
expertise and experience like I don't if it's a tiny startup working in
maybe retail or like a lower risk application, then, you know,
I don't think they should be subject to the same requirements and regulations.
So I think it really has to have a application based approach to
transparency requirements. Yeah.
Super interesting. Chris, what do you think about this?
I mean,
I think what I'm hearing from Kate is almost like kind of this proposal that,
you know, Anthropic is very much like, look how how big the model is.
That's like kind of the relevant place where we draw the line.
I guess here is almost kind of alternative, which is we
almost should think about this in terms of like use case.
That's actually where the rub is on some of this stuff.
I think if the only thing that is stopping you developing a chemical weapon
is closed system prompt, I think we've got a lot more work
to do in the industry that that would be my starting point now.
I mean, a more a more serious point or I
for a second, I, I understand where it's come from.
And I do think anthropic, as Kate said, is one of the most responsible companies.
And, and I think they've been
very open in their papers and, and they do a lot of work on safety.
And they talk about it openly. And I think that's great.
But I just fundamentally disagree.
I don't I don't believe in closed models. Right.
I use them, obviously.
But I, I really think the most way, the best way of being transparent
is publish what data was used to train your model and publish your weights.
And we can all have a good look at it.
And then we can tell you whether it's a good model or a bad model.
And and you can improve it.
But what if it's locked behind the door?
Then how do we know?
And it's like would if you were buying, I mean, to take your example, right?
If I went into a store and said, can I have a soft drink in there?
And it said, don't worry, you won't die from drinking this.
Trust me, it's got secret ingredient X, you know, like a sales pitch.
I'd be like,
maybe I'll
buy the water over there instead, at least, like, you know, I.
And I think you just gotta, I, I get the point
about labeling, but, I mean, come on,
you know, the problem with labeling and, you know, what goes on in those areas.
So I think it comes from a good place.
But but I think transparency ultimately comes from being more open.
And therefore if everybody is more open, I think we're going to get to this
sort of issue of being able to understand how to fix
all the bad problems with a wider set of eyes, being able to look at that.
Aaron, if you'll let me play, I don't even know if this is Dario's
position, but like if you let me play like devil's advocate for a second,
you know, I think company like Anthropic might say, well, the problem
with what Chris is proposing is
it would just give all of these dangerous capabilities to all sorts of people.
And that's even more dangerous than like, what we're proposing.
Do you buy that?
Like kind of that actually does these risks with open this on AI,
where theoretically we'd
have more transparency, but it might not necessarily guarantee more safety?
Yeah.
I mean, I mean, this is a very complex problem, right.
Yeah.
And so and so I don't think, I think that there's
a, you know, one shoe fits all, you know, you know, solution here.
And one of my biggest issues,
that really made my skin crawl, right, right around this was this selective
transparency, right where we can create this unfair competitive barrier.
And it seems counter to what I think,
you know, perplexity here and anthropic here is,
is wanting to do, you know, so, I mean, what is a frontier model, right?
I mean, sometimes a frontier model.
It started by these startups, small companies,
you know, they're the ones that begin this.
But on the other hand, there's a lot of loopholes here
where we're where we're limiting the application, right,
to just the largest developers, where if you have
something like 100 million of revenue or 1 billion of annual capital
expenditure to be liable, right to follow these minimum standards.
But frontier models, again, you know, you know, tend to come out of startups
a lot of the times,
not all the times, you know, and, and then and then we're saying, okay,
so the most powerful or these advanced AI technologies, you know,
have a loophole where they don't they don't need to follow this.
Right.
Only until a big company either acquires the model or,
you know, like OpenAI becomes a big enough company
to have to, you know, you know, bow down and then follow this.
I mean, it is very ambitious
what they're proposing, because this requires, I think, laws.
Right.
So lots of legislation has to go through, right, to make this happen.
There has to be a comprehensive list of evaluation
methods, you know, so but again, it's it's a good start.
Right.
And, you know, I, you know, if you even look back right in the past,
what was it
July 2023 where OpenAI, Google, Microsoft and so forth,
they formed this frontier model forum, right?
Right, right.
And we're still trying to get to these AI safety, you know, protocols.
Right. And summits and I think,
this Anthropic release of this right, you know, helps to kind of push it forward.
But let's just be careful about this selective transparency that I think
might be inherent here and not be counter to what's actually happening
right in the field.
I wonder, though,
if we're, like, letting you know, perfect be the enemy of good here.
Like, of course, open weights are more transparent.
And I completely agree, Chris is the way to develop responsible
AI, but open AI is not going to just open up overnight.
Anthropic like these companies have far too much commercial value
behind having their proprietary weights just all of a sudden
put things in the open.
So, you know, I see what Anthropic doing here is maybe
trying to create what is some near-term wins around
making sure we have basic standards and transparency and a call to action.
I think again, they're they're approaching it a bit of the wrong way.
And I agree with the selective transparency issue, Aaron.
I think the transparency needed needs to be guided by the downstream risk,
not by the some arbitrary definition of the size of the model
or the revenue of the model provider, that type of thing.
But, you know, this is ultimately a good thing.
It's better to have some transparency, see the no transparency, right.
When it comes to frontier model providers.
But even the even if you don't open up the weights and
and I'm a big fan of open it the weights if you say this is how we train the model
and this is the data that we use from transparency, the reality is
there's only a certain number of companies
that are going to be able
to take the amount of data and have the compute to be able
to train that model in the first place in that sense, and,
and ensure they can get rid of some of the algorithms or whatever.
But actually ultimately what goes into the model is important, right?
Because the reality is, is
and again, I don't know what
they're feeding the model, but if if you got 20,000 pages
of how to make a chemical bomb going into the model, well,
guess what the model's going to learn is going to learn how to do that.
Right?
So maybe we can have a discussion of what goes into the model in the first place.
So but at the moment,
you know, transparency wise they're not being transparent about that.
So and I get it model safety cards.
And I guess lots of companies are not being transparent
because they think that's their secret sauce.
But if we want true transparency it's tell us well why name.
Yeah.
And so there are different degrees of transparency right.
You can and I don't think at least at this point
they're saying we have defines what transparent means
or this is exactly what thou shalt do to be called transparent.
There's a lot of great frameworks out there.
Like Stanford has a transparency index that they they run for different model
providers, and it could involve sharing the data.
And I completely agree. That's super important.
That's why we share all of the data sources behind how we train Granite.
And we're really open about what data goes into our models,
to better help enrich conversations around the model itself
and performance and safety and all of those things.
But there's also a lot of, like, real evaluations around safety,
around bias, understanding the limitations of the model that I think
are absent from the discussion.
And even that is, you know, a degree of transparency that is missing
from some of the frontier models today that I think they're calling for,
which seems like a reasonable place to start.
Yeah, just a real quick point.
Wanted to just chime in, too.
You know, we've been talking about the the organizations and, you know, the,
companies and so on and so forth that creating this, technology.
But I think some of the onus has to come on the consumers or the users, right?
That that whenever they decide to use a particular model,
they need to agree and abide by certain terms and conditions.
And if they don't,
then some of these minimum standards here that are here, such as the guardrails,
can detect and automatically flag somebody's usage right?
As nefarious, maybe. Right.
And then and then do something about it. Right.
So so I think it's on both sides.
It's the producers and the consumers right, of these tools.
And perhaps Anthropic could take a, a view.
Right. On that. Right.
And then I would also encourage Anthropic again, you know, that selective
transparency and giving a loophole for the small companies
might be a dangerous thing.
Okay.
If we can end the segment with you on just a really practical note,
because I feel like you're really well
positioned on this issue because you're like,
you're in the trenches, you're in the arena on this.
Like, how is Granite, I guess, thinking about transparency.
How does it think about model cards,
you know, kind of these this whole set of issues, it's
obviously a little bit different from the anthropic setting,
but I think it people benefit a lot just for hearing about
how the team is sort of thinking about through these issues.
Yeah, I think anthropic, it's right, is recognizing that safety
is such an evolving field and really not
well understood to the degree, today there's so much left to continue to work
on, that they're not trying to prescribe what safety is.
They're just talking about transparency.
And that's very much the approach we're taking with Granite.
So trying to recognize that this field is evolving so quickly,
the best way that we can arm our customers, arm our users,
ARM developers is to be as open as possible.
While, you know, maintaining responsible use of and stewardship
of any data that we have curated and used for training.
So where again, in terms of our high level lieutenants,
we have a very rigorous data governance and review process for all the data
that we create, such that we are very open of every data set and source
that's used to train Granite.
And we share all of that in our technical papers.
We also have, you know, robust safety framework.
But again, safety is something that is evolving.
So I think all you can do at this point is just try and be as open as possible,
and share where they are, where there are known risks,
where there are unknown risks that we're still,
working as a field to better understand and figure out how to measure
what mitigations are recommended and everything else.
That's one of the reasons why we also include tools like Granite Guardian
alongside our Granite models to help better manage risks,
with AI model deployments, knowing that not everything
can be baked at the most fundamental level into the model weights themselves.
And then, you know, trying to be, open about how the model is distributed
so that the community can use it, test it, kick the tires,
give feedback, tell us how it's going, and also report issues.
So all of our models are distributed openly under Apache 2.0 license
on Hugging Face.
In addition to making them available in our products, we have a security
incident reporting program where you can report issues.
We're actually about to start a white hat hacking program,
and bug bounty program for granite, which I'm really excited about.
So more to come there.
And so really, we're working with a lot of different partners
that, you know, we'll be sharing more about in the next couple
of weeks around safety for granite, trying to involve the community
not just on using granite, but helping to report issues and make it better.
Yeah. That's great.
We'll have to have you back on the show when you do the White Hat Bounty.
That'll be really, really cool.
I all right, I'm gonna move us on to our next topic.
Obviously a lot to cover there and we'll come back to that topic in the future.
Interesting story popped up.
Cloudflare, which is most known for being at least at this point, I would say
part of the core infrastructure for the internet,
largely started as a platform for managing spikes in traffic.
And I think is is widely known, I think, across the web, as is sort
of the protection that a lot of websites use against, say, DDoS attacks.
And there's a really super interesting thing that came out
where Cloudflare recently said one of the problems
we're noticing across a lot of the websites that we protect
is the the traffic and the load that's coming from AI scrapers
going all across the web to bring down data for training purposes.
And so what we are going to do
is that we have a, system for blocking these crawlers.
And rather than having to have websites opt
into it, we're going to start blocking those crawlers by default.
And so this is a really big deal
for a lot of the companies that have been relying on these crawlers
to like acquire data for training or fine tuning purposes.
And I think has kind of sparked a big controversy over what
the norms around data should look like, because I think Cloudflare now, in turn,
is saying, well, now that we have
all of these groups of websites and publishers that we are protecting,
you know, our our ultimate mission may be to think a little bit about how we
monetize, right?
How do we basically ensure
that both Cloudflare gets a benefit and also these websites get a benefit
and that these scrapers can't just sort of scrape the value, quote
unquote, out of these websites, without kind of the consent.
Of, of the publishers.
And so I guess Aaron, maybe I'll kick it to you.
I mean, simply stated, like, is this a good development for the web?
Does it like, solve the problem of eye scrapers once and for all?
Well, so.
So I think this is a fascinating story, right?
As it's evolving, you know, speaking and speaking of evolving,
you know, the internet, right, is evolving.
You know, we just talked about the AI based browsers.
I think we have AI based internet, right?
We have new protocols that are being created like ACP, which the
agent communication protocol agent agent, you have MCP, the model context protocol.
Right. So so those set of protocols
were creating this to me, this this new network right of agents,
you know, where we have and and and and if you think about it, agents
could become like the new websites, right where you go visit an agent, it
on demand creates maybe a website or creates data that's available.
And so traffic to agents now garners
payment rather than traffic to websites that garner, garner's payments.
So so I think we're in that paradigm shift where we're moving towards,
you know, that, and and so this so Cloudflare
enforcing this permission based model for the internet is sort of,
this, this struggle, right, between aiming to give content creators
more control over the content, which to me, and the longer term,
I won't be as relevant per se,
just just because we're going towards this AI internet.
But it could also be bad for how AI models are trained and use.
Like this In-context learning and access to these large data sets.
And so I'm very excited
to see, you know, where the this this whole internet does go.
We just need to be careful that we don't restricted right and
and I do think perhaps, a decentralized, way.
Right where, for example, by using MCP.
Right.
By by by using a tool,
that could pull data from your own private repository
that isn't publicly available on the internet.
Therefore, you control your own content as a, as the content creator,
and you could charge someone by visiting your agent, you know, piece.
So, you know, you know, I think there's a balance here.
Right.
And and I think we'll quickly get to that kind of balance where we're going.
I mean, it feels a little bit like we might be seeing
a tragedy of commons in the practice in process.
Like I think like one of the reasons the internet was open
was sort of the expectation that you wouldn't
be like driving a semi-truck through the front door.
And I think maybe now as kind of like agents and crawlers become more
and more active, it's not as easy to maintain that openness or as costless.
It's not as cost us to maintain that openness.
And I guess this is like in some ways a very natural response.
But I feel like I have like a little bit of,
I feel a little nostalgic about our a little sentimental about it.
Right. Because it's almost like an enclosure on the web.
I don't know if you feel the same way.
No, I, I completely agree.
I mean, I, I feel like a lot of the promise of the internet
and a lot of the value that I don't know that people quite realize how much value
just the society as a whole has gotten from being able to crawl
and keep snapshots of the internet for others to use beyond even just models
like look at how search has evolved like we have now very detailed
historical snapshots of the internet, which is basically snapshots of society
going back a couple decades thanks to crawling.
And I do think it's this common resource that has this big opportunity for master
democratization of information that now, very aptly put, a tragedy of the commons.
We're going to start to see erode.
And I understand why, to your point.
Like there's a lot of, frankly, bad actors like this should be a symbiotic
ecosystem of mass sharing knowledge and building products off of that knowledge
that make it more accessible, that all can benefit from it.
But we get reports all the time of different model providers.
For example, ignoring robots.txt crawling information
that is, you know, pirated just so that they can use it for training.
Like there's all sorts of these like bad practices
when you then compile it together and then also look at the fact
that they're resulting in these incredibly, you know,
commercially attractive opportunities that are very one sided.
That value is not being distributed.
It makes a lot of sense why content creators are saying,
okay, this is no longer, you know, benefiting us as a whole.
We need to start to, to put up some fences.
And, you know, I think it's a tricky position that we're in.
I get why they're doing it.
I think we as a whole, you know, as the industry, it's kind of abused
this resource, because there are these irresponsible actors all operate that way.
And so now we're we're going to start losing,
a very tremendously valuable resource
that could do a lot of good if it were fully open.
And I think the only way out of it is we need to get to a much healthier
discourse between content providers, organizations like Common Crawl,
who are, you know, helping create these repositories of the internet
and model providers and figure out how we can get to a more
responsible approach with more, you know, distributed, distribution
of the benefits that are being gained from this information.
Yeah. For sure.
Chris, I, I guess I'll admit to being, I mean, as part of the story,
feeling like a little creeped out.
Right?
Like in some sense, like Cloudflare is like we're standing up for the little guy.
We're standing up for the little publisher.
But it also seems like if they kind of are able to accomplish their vision,
they end up controlling large swaths of the internet as well.
Right?
They get to decide who gets to scrape, how much money they get paid,
all that kind of stuff.
And I think probably someone
like the CEO of Cloudflare would probably say something
like, okay, well, what's your solution, smart guy?
And I don't know.
I mean, should should we be sort of creeped out by this development?
Because I think on one hand,
I think it's very easy, I think, to cheer for what Cloudflare is trying to do.
On the other hand, it also seems to lead us down to a path
that that might not also be all that great either.
I don't know, I mean, this is probably the bit that gets edited that anyway.
I mean, it was a great segment, but it's not going to appear on the show.
I mean, take the chat like,
I mean, I think the easiest way to keep it in is to take the other side, right?
Which is like, okay, well then I mean, what's the alternative?
Right. Like what's what's the idea now?
I think it's hard, right?
Because the really is it's you need big players versus big players, right.
So to have that level of protection.
But I, I hope that this is a, transient thing, if I'm honest, because I,
I honestly think the, the internet
as it stands today is designed for human being consumptions and for the browser.
And I think they are not designed for AI.
Back to Aaron's point, right?
Protocols such as 2, MCP, etc.
is really about trying to understand what that what those protocols look like.
And if we think about headless content, for example, most organizations
are hosting their content and content management systems
in a headless way, so they're not storing it as HTML.
What they're doing is storing the content as structured data,
and then they're mashing it together
with a template and rendering it out into HTML.
That is pretty I mean, you know, you will have frameworks like react
and all these other things in the way as well.
But, but, but the reality is the source content is not stored
at in that rendered format the browsers expect.
So what I think is going to happen in as we change through,
into the future is that we will start adaptively
rendering our content to the types
of, bots, etc., that will be hitting at it.
And therefore, you know, we know today that bots like markdown,
so maybe they're going to get markdown in that case, and then maybe new formats
which will be more appropriate for training and will come.
And therefore at that point, in the same way as HTML was designed
as a markup language with controls, and you're going to be able
to start putting in things like saying,
well, actually, if you want to be able to pay for that content for usage,
please do this.
So I, I,
I think this is a transient thing, and I think there are probably better
content protocols that will be designed and can be designed over time
to deal with these problems.
But just now is a bit of a free for all I think is great.
Honestly, I will take the other position.
I think it's great that Cloudflare is standing up for that,
because actually it's something that companies that have genuine,
you know, content
that is they've curated, they pay good money to create, etc.
as valuable content is, is being taken and then they they're not getting the money.
The monetization from that
because they're not even getting the clicks coming through,
from, from Google etc. any more.
So I, I sort of understand that you have to do that monetization,
but I, but the flip the me all for suddenly
the Savior comes down and then padlocks my shop up
and tells me tells you who's allowed to come in and who's not.
I'm kind of like, I'd, I don't know, I you might
I want a little fire there but but I think it's done with good intention.
But then yeah you.
Know we'll see.
Kate, I'm going to ask you to do a little bit of wild speculation
for the list before we close out this segment.
I think putting two and two together, it seems to me that we're about
to get into this big clash where there's going to be these AI browsers
that are going to basically try to negotiate deals to access
content, you know, from publishers online.
At the same time, you'll have folks like Cloudflare,
like trying to create like a shield for the publishers and saying,
hey, we and there's almost like a little bit of a race between
who will control the browser and their access to content.
And like all of these kind of more interstitial applications
like Cloudflare that are also trying to get a piece of this.
Do you have any theory for like who who wins in that race?
It's like,
I mean, we're at the beginning, very beginning of both,
but it's kind of like, it's kind of interesting question.
Yeah.
I mean, I think it's going to be a really dynamic period in the, this next
evolution of browser wars with all these different factors coming together.
You know,
I think in is if we look at like, traditional platform battles, right, it's
all going to be about who can create that value opportunity that attracts users,
that then creates that snowball effect where their content providers,
in order to get clicks, need to then get on that browser,
and then all of a sudden all the dominoes start to start to fall.
So, you know, I think there's going to be
maybe some interesting like first mover advantages here.
If somebody can strike up a really good deal with OpenAI
or someone else, Perplexity on their browsers for top content,
making their content accessible and with paired like importantly, paired
with real value creation from the the agents or models
that, these companies are embedding behind the scenes.
You know, whoever can kind of land some really killer use cases
around both that content and the models working together.
That just make this.
Everyone just has to use it. It's too compelling.
You know, I think it's going to help them set that the series of dominoes
to start falling, so to speak. It's very interesting.
All right.
I'm going to move us on to our last and final topic.
Aaron, I hate typecasting you,
but I feel like every time you're on we always talk sports.
But, it is Wimbledon.
You've been doing a bunch of stuff around Wimbledon.
I think you've been there.
What's, What what are you working on this year?
Yeah.
So, I mean, so first of all, I'm very proud of our work and our team, right.
Because we took on this grand challenge, right, of creating this real time
and accurate.
So it's a large scale system where fans could interact with.
So it's core consumer facing.
Right.
And this was all about creating, you know, this experience for both speed
and accuracy around, you know, two main applied R&D projects that we had
which I'll get into in a minute.
But but I will say, you know, the A championship is going on now.
So if you go on with wimbledon.com, you could go on the mobile app,
but you can see the see there.
Our work, you know, right now and you can use it what we call match chat.
Right.
And then see our live likelihood to win estimations.
There.
But what, what I'm most proud about right, is, is how we came together, right?
And created this, Match Chat where we worked with Wimbledon.
Really?
To reimagine what it means to follow the game or the match.
And this is where we debuted it.
Match chat, you know, just a few weeks ago.
And, it's this interactive, real time, what we call AI assistant that's built
on top of these, a genetic, we use line graph, you know, so
it's these graphs of agents that all work together, to answer fans questions.
Right.
And and it's and it's really sophisticated what we did.
I can go go into the details, but I'll just mention that,
that there's many different, edges.
Right.
And nodes where all these agents could, could, could work together.
And because it has to be real time, right, with this very large scale,
put in a lot of these, threads where they had timeouts.
So if it didn't respond within a certain amount of time, we would stop,
and then we would go back into a different type of agent that could respond faster.
Right.
And these agents would run in parallel, such that we would have, you know, this,
this pull in the tent that could answer, you know, within a reasonable
amount of time, given
all of the different types of questions that you could have and so on.
So, so that's, that's that part.
And now the other part that I'm very proud about is our live like likelihood to win
where, you know, where as a match is progressing.
Right.
We show this with we tell the story of the match
about who we think is going to win based upon many different factors.
You know, so we look at the performance of play.
Right. So so it's real time. It's streaming.
And so as someone hits the ball we get that information right.
So so how many aces what kind of for him what kind of bat can and so on.
We we we look at the score.
We have different decay boosters and factors that we put together.
So we built this series of equations right.
That help the fans to understand what's happening.
And it's and it's all built, predicated upon what's called our pre-match
likelihood to win, which is,
an SVM, series of models.
Right. As long as decision tree.
So it's traditional predictive modeling, right.
There that that then gives us all of this.
So, so it's pretty cool, you know, you know, as these are combined together, one.
Question that I had for you on this is, I think last year
we were talking about this, and I think it's just great to check
in, you know, every time you come back because I think this is kind of evolving,
you know, as part of what you and I think your team is working on is like
sort of architecting or reimagining what the fan experience should be
for your spectator watching the game.
Right? Like, what are the things that you expect?
And a lot of these tools are like very different to,
I mean, I guess the old school where you watch tennis,
which is, I guess you like, sit and then you like your head
pivots back and forth and then like the game is over.
Do you feel like, like fans are adopting these tools in a big way?
Like,
do you think, like, there's certain segments of fans that are more into it
than others just kind of curious about, like, adoption? Yeah.
So so we have a paper that was, published ACM Kddi on Barcelona.
Right. And, you can access it.
It's, actually open source paper.
So if you just search for it, you know, let's just say if you Google it, right.
For example, you can find it and read about it, you know, it's,
it's our I mean, we've been doing this for about, what, five years, right.
And and just, you know, got to get to put a plug in for, for our team.
And, and when the IBM corporate word right around this work,
but anyhow, our, our fans are adopting this.
Absolutely right.
Fans are not, not not only adopting it, but they expect, you know,
this type of work, which is about challenging,
but it's also very fun and interesting, right, to, to innovate together.
At the beginning, you know, we were looking at a very broad,
you know, fan base where, you know, let's not be a, a toy store for.
All right, let's, let's select selectively pick a couple of the personas
that we want to focus in on and be very good at just that.
Right.
And and so that's where this AI commentary really started.
Right. Where, where we would do, you know,
go through the series of pipelines
to create commutation and then put them into these AI video highlights.
Right. And, and
and it was very high level, you know, so it was meant to be for the generalist.
But now if we fast forward to today with this match chat, you know, this this
you can ask virtually any kind of question you know, that you would want right.
So it's we handle, you know, the, the the extreme fan who wants to know
in detail every single state that that that that's happening.
Why is somebody supposed to win.
What's the biggest factor
all the way down to somebody who wants to go eat a strawberry?
Right, right, right on site.
So, so it's a whole gamut and, and it, it's going to continue to, to evolve.
Yeah. That's great.
Okay. Last question for
you is it feels like a lot of what Aaron's explaining is kind of like,
I mean, one of the great things these models are good for is prediction.
And of course, one of the great activities of any sports fan is prediction, right?
It's like, who's going to win?
Who's up, who's down.
Is there kind of I don't know if you think there's like, a worry in your mind.
I just kind of curious about how you would
speculate about
this is like, if our prediction gets really good,
is it going to take a lot of the thrill out of sports?
Like we just know who's going to win before the game starts?
I think it just raises the stakes.
Like if we get better and better to predicting
and then someone beats the expectation, that's like even more, even more hype.
Yeah, yeah, even more hype.
I think it is interesting consequences on the gambling industry.
That's probably a whole different episode. But,
you know, I think, yeah, I think we're going to keep evolving.
We've seen a lot of great examples.
Throughout time, most recently now with Wimbledon, of how I can augment
our understanding of the game.
And part of that is improving predictions.
Part of that's just predicting small parts of the game, like where does
where where should you look like if you're not sure
where the ball's about to go or the puck's about to go,
like it can help guide you and better understand.
And so I think there's also a lot
of educational aspects that deal with prediction that aren't just like,
you know, so-and-so has a x percent chance of winning this match.
And then there's a big upset.
Yeah, yeah.
It's funny too, because we had noticed, at times
when players look at the predictions, they actually change their playing style.
So it's the chicken or the egg, right?
So we, we make a prediction that a player changes their style.
So great power, great responsibility right there. So,
Chris, maybe I'll let you wrap us up.
I just think about a conversation I had a friend with, with a friend
a number of years ago who is like,
I've been getting into baseball recently, and I was like, oh, that's interesting.
You've never been into any sport before.
And he's like, yeah, this is like a lot of really good numbers and statistics.
And this is like, why I'm really excited about this sport now.
And we've been talking a little bit like in Aaron's narrative about how
this is like, how do we take tennis fans and enhance their experience using AI?
I think the other thing that this sort of leaves me with
is whether or not like, it's
going to work the other way, like if you're really into AI,
whether or not stuff like this will pull you into sports in a way that you haven't,
and whether or not this actually opens up in some ways like new markets
for, for, you know, sports spectating that maybe haven't existed before.
Yeah, I think so.
I think fans get into lots of different sports for different reasons.
And I think, for some people that's going to be data, right?
And prediction and some of it's going to be, what the strategy is.
So I think that, for people
who are more technically oriented, I think it kind of opens up the game
and allows you to see it in a different way.
And I actually think that's really good for hardcore fans as well.
Right.
To to Kate's earlier point about where the ball going to go, etc., you start to
you start to think a little bit more and and so I think it's totally fascinating.
I mean, I remember once I, I hate to say this and I scraped
all of the IPL cricket from India once.
And I, I've only went to one game, but, but it was fascinating.
So I pulled all of the data in there, and, and then I remembered
the kind of the Malcolm Gladwell thing about, you know, about ages of, you know,
what month school cutoffs and things like that, and how big were people, etc..
And, and so I ran the data against the IPL cricketers from India
at that point, and I and I and I found the exact same cutoffs.
Right.
So, so their school year started
in September, but, but, but pretty much, if you were born
and there was like nobody in June playing cricket in India, just nobody.
But then, you know, you think, oh, wow. Okay.
This is this is, you know, this is the kind of, case there was,
you know, the game is sort of solved in that sense.
But, you know, the earlier that you were born,
the bigger that you are, you know, the more you win.
But then like Sachin Tendulkar comes across and you look at his stats
and when he was born and it's just like blows everything out of the window, right?
So, now I know nothing about cricket.
I'm sure lots of people are going to complain.
But but it found me.
I got interested in it for a couple of weeks. Right.
Because I was interested in that stats and data and,
and again, when I think about things like football
stats and data is it's just such a big thing, regarding that.
So I just think it brings new dimensions to the game.
And, and I think it's, it's going to get more people interested in these games,
and I think it will change the game as, as well. Right.
And start to bring elements to, to, to increase engagement.
So I think it's a good thing.
Kate, Chris, Aaron.
This is one of my favorite panels for MoE
So thank you all for coming back on the show.
We'll have this exact panel on, in the near future.
I'm certain. And thanks for all your listeners.
If you enjoyed what you heard, you can get us on Apple Podcasts, Spotify,
and podcast platforms everywhere,
and we'll see you next week on Mixture of Experts.