LXA CEO Carlos Doughty and Godfather of Martech Scott Brinker recently sat to discuss a range of topics, with the rapid growth and potential of AI being the major theme running through the discussion.
In part one, Carlos and Scott discussed the growth and massive potential of AI. In this part of the discussion, Carlos and Scott discuss the potential winners and losers around AI, why data has the potential to be the deciding factor, and why Microsoft is way ahead of Google right now.
Carlos Doughty: I think an interesting question is around the commoditisation of AI - what does it become if we've all got the same access to the same smart intelligence? How differentiated are different vendors? Where are the real winners or losers? Is it just a case of being so unbelievably niche and use case specific that you wrap around something better than everybody else?
Take ChatGBT. Is it just too open, covering everything? Is there too much data and do you want curated data instead? What do you think?
Scott Brinker: I am so excited about this on multiple levels.I think most people would probably now agree, the answer is the data.Everyone's gonna have the same AI engines, so those are effectively commoditized. Certainly, open data out on the web is commoditized now too, so where's the differentiation?
The differentiation is from companies that have unique datasets, and proprietary data sets, and then they're able to build ecosystem relationships with other companies to combine datasets in ways that help that ecosystem to have an incredible advantage.
“In every disruption, it pits the startups against the incumbents. In the Generative AI disruption, I think the incumbents are moving faster than I would have expected and they have a ‘data advantage’.”
I saw a tweet from Brian Halligan which is very relevant to this. In the history of disruption, It has usually been the disruptors that have an advantage over the incumbents. It's the classic Clayton Christensen's Innovator's Dilemma.
That might not be the case this time around because of the importance of data for AI models. It's the large incumbents that mostly have the largest proprietary data sets and these existing ecosystems. So if they can leverage that data, using these new open AI engines, then the incumbents have a chance. They still need to be fighting for it, but the incumbents have a chance to defend their position because, for a lot of startups, that's going to be the piece they're missing.
"Are there going to be some businesses which completely transform who and what they are based on data they've been collecting, that hadn't really been monetised and turned into a business model?”
CD: I agree completely. There's a term ‘data sawdust’ I love, which I heard on a podcast. It was from the CEO/ Founder of CB Insights (Anand Sanwal).
The concept he was talking about was - are you sitting on gold? There’s this stuff that you pay no attention to that’s just sitting there. Is it worth more than you think?
I think there's also one interesting thing - are there going to be some businesses which completely transform who and what they are based on a bunch of data they've been collecting, that didn't really mean anything at the time, or they hadn't really monetised and turned into a business model.
This relates to a concept I’ve been exploring recently, the AI of the possible. If you look at some of the big beasts and the troves of data they're sat on. If you look at Amazon, and all of the data and insights they have from Alexa, Twitch, Prime, and MGM Studios.
None of those businesses was set up with the idea of collecting proprietary data that can be used for AI purposes in some way, shape, or form. Whether it's Meta with WhatsApp and Messenger or it's Google Mail.
Clearly, there are privacy and ethics issues attached to this data and how it's used. Let's assume there's a way around this where it's aggregated and anonymous in a way that means there are no privacy issues.
This data then provides some insights and natural language understanding that's unique and truly proprietary. For example, the conversation we have on WhatsApp is different to the conversation we have on email, and the conversation we have on a call is different from a presentation.
I'd be really interested to see how some of this plays out, and how that data is used or not used. Is it going to be licenced out? Or is it going to be retained and used to create?
Right now we're all amazed by OpenAI, backed by Microsoft. And they’re 10 years ahead, but is that really going to be the case or will we end up with two or three major players? What do you think? Do you think the big tech companies are going to make use of some of this data?
SB: The one who I feel has the best opportunity at the moment is Microsoft, and I have to say I've been a fan of Satya Nadella. I think the turnaround he created in Microsoft was already astounding - the rise of Azure, this Open AI deal, to have the prescience to make that deal the way they did when they did was a brilliant move.
The way that they are incorporating this into the Microsoft universe, so that every Office document, Excel, stuff, everything in dynamics, everything we could have in Bing, everything we could have in Power BI now is all connected.
And so much of the data in any one company flows through the Microsoft ecosystem somehow. For Microsoft, there's what they might do for this data for their own purposes. But let's even set that aside.
Let's say all they do is applied to the data in your business that they're uniquely poised to have all this visibility on. I think Microsoft has set itself up where out of the big five I think they've won because that's the piece that's going to be really, really hard for any of the others to replicate.
I mean, Meta has a slice of your world, but it’s only seeing a small part compared to Microsoft.
CD: Microsoft has got the business model ready to roll and that AI layer just absolutely breaks them away. Whether it's Teams on steroids, Excel on steroids, or whatever.
SB: We are on Teams right now. You’ve set up Otter.ai to record this, but actually, you soon won't have to do that, right? You'll just say, hey Teams, could you summarise a blog post for me and put it in a Word doc? Or, I'd like an Excel spreadsheet of all the names and URLs that Scott mentioned.
“Google is in serious trouble. We'll see what they come up with because they clearly know they're in serious trouble. Google's mission is to organise the world's information and make it universally accessible and useful. They just got owned by ChatGPT. Because that's exactly what it is doing.”
CD: Or we want to do a follow-up chat, so find a time that suits both of us and book it. You forget like that's where they make their money already. They've got a dominant market share, plus they absolutely blew away Slack, which they really cut off at the knees.
Google has just been left alone for years to have 95% of the search market and they don't need to take much off them for that to just transform their business, right?
SB: Yeah, Google is in serious trouble. We'll see what they come up with because they clearly know they're in serious trouble.
I was just looking up Google's mission. Remember this mission statement? Google's mission is to organise the world's information and make it universally accessible and useful. They just got owned by ChatGPT. Because that's exactly what it is doing.
SB: The edge for any business now is their data, their unique data, their unique content. So you take all the talks from AnticonLX for example, this is actually a really interesting, proprietary data set.
What are all the things we could do? Have you ever wanted to tap into the knowledge of, you know, the best marketing minds in the world?
CD: I think the 45 terabytes of data is nice with OpenAi, but maybe there are just five sources you really care about - this Twitter stream, these three books, this person's blog etc.
It's a bit like a Google search. Google decided at some point somehow, that this was the truth, that this source of information is the truth and it's the best version of it and clearly, they did it in a relatively smart and intelligent way.
At the same time, they had this God-like power to control what you see. ChatGBT is a new show in town for the same thing. You don't really understand how this black box has arrived at these answers, it's probably a sort of generic version or a fast, smart, organised part of it.
We might not really want the generic version, or the most common one. This is the reason why you sign up for certain newsletters, go to certain events, or listen to certain people because their brain is interesting, they're smart, and they see something everyone else does.
It's an interesting thing because one of the things I've been pushing is our learning experience platform, which currently doesn't have this capability. I'm hoping Ai is going to bring it.
Instead of the old search filters, I want people to be able to say, for example, I want information on marketing operations. And I’d quite like it to be about lead routing and scoring, and it would pull the answers for you, but it would also suggest a video you should watch and it's not the full 30 minutes, but a useful segment of 17 minutes in that it chops out for you.
The ability to curate and aggregate is the gold, right? If you have the trusted source of information that you put the most value on and then the ability to apply smart intelligence on top to chop it up and give you what you want.
This was the second of a three-part interview with Scott Brinker. See the other parts of the series here: