Get into the gAI protection business
I’m Sven writing this to help you build things with data for data PMs, startups, investors, and leaders.
New market: protection against gAI theft
Databricks buys Okera for the right reasons.
OSS is NOT the future of AI
Subscribe (free!) or someone will steal your data business & (data) users.
(1) Get into the gAI protection business
First, DALL-E steals your painting style, and now ChatGPT and his gang steal your writing style.
Luckily for all generative humans out there, research is starting to act. It’s called Glaze, and it’s the first research-driven “anti-generative AI” tool I’m aware of.
Glace alters your imagery work slightly in a way that makes it impossible for generative AIs to train off your work.
Here’s what you need to know:
People are scared as shit about getting their content stolen by AI.
Companies like Reddit are taking action to prevent content stealing.
Research is picking up on the theft problem.
The good stuff for all company founders & PMs:
There is now an emerging market for gAI protection!
It’s driven by fear. And strong emotions make for easy sales.
The protection solutions will have to evolve in parallel to gAIs, the Glace creators say.
The gAI market will propel this market forward and will keep growing.
Now someone needs to show me the tool that inserts weird ASCII characters all over my content to protect me from the gAI gang.
(2) Yet another AI sell
Databricks, the lakehouse, and AI company is buying Okera, an AI-powered data governance solution.
The crazy part? The sell is NOT about AI.
The acquisition makes sense and is an investment into the future of data.
Okera does some AI stuff to classify data, e.g. into PII or non-PII.
But Okera also manages access, and that’s key!
5 years ago, everyone accessing data did it through SQL.
SQL has excellent access management.
Today, we lost SQL as the central access point.
And that trend is strong; data access and storage will continue to be increasingly decentralized.
So we need a central solution to manage decentralized data access and storage.
Okera does just that.
Of course, the most pressing issue that everyone needs a solution to right now is the identification and access management for PII. That’s the prime use case for Okera already.
(3) OSS is NOT the future of AI
Tristan Handy, CEO of DbtLabs, and other people are starting to think OSS might have a chance in the AI race.
I love open source, but this isn’t going to happen—quite the opposite. The ones who see beyond “I want this to be true” will have a significant advantage over those who don’t.
Here are two facts you need to know to make wise business decisions regarding AI.
1) Training data, AI-model users, and computing power are all essential to create great AI models.
And all three of those ingredients are dominated by network effects. If you have more training data, your existing data becomes more valuable, and so on.
Network effects always lead to winner takes all situations. The company with the most training data, AI-model users, and computing power will attract 90%+ of the total market.
“OSS models will always be 3 years behind closed models, and that’s a good thing“ (Sam Altman)
2) Companies like Google are exceptional at using OS to disintegrate markets.
Google is a massive player in the Cloud market in part thanks to the success of Kubernetes. They managed to disrupt the mobile phone market with Android.
Notice something? These are all open-source solutions. They all made Google much richer and enabled it to capture complete markets.
I have no doubt Google will find a smart way to open source the parts of AI models that enable Google to dominate the entire market.
So what do we do?
Don’t try to compete with Google.
Don’t try to compete with Google.
Expect serious infrastructure to make building AI model-powered products easier to emerge very soon!
Don’t try to compete with that, either. Instead, integrate it into your products.
Share Three Data Point Thursday
Recommend the Three Data Point Thursday to anyone and get the 20 Point Questionnaire To Assess The Strength Of Your Data Startup Idea.
Just share the ThDPTh on Twitter, LinkedIn, or wherever you got some space, and respond to this email with “SHARE: [link to your recommendation],” and you’ll receive this fantastic giveaway.