Bullish:

• 🟢 $SENTIENT - Discussing the potential of Sentient's AI x crypto projects, its $85M fund raise, and emphasizing the importance of decentralized AI and participatory economy.
  • “So Sentient is a model company, and we'll build our models and give it to others. And if you like what you do, you can use a Sentient protocol to do the same thing for yourself. ” - Himanshu (0:00)
  • “Maybe one way to think of it is the GTM strategy that we want to be for the initial phase. We want to be the company who's providing and building the models because because we want to as I said, the thing with AI is it's a working feature. No matter what what I tell you and what I do, you first will compare me on quality with OpenAI. Maybe you'll give me some elbow room, some slack. But then at some point, you will say, dude, I like, look at this thing.” - Himanshu (0:00)
  • “And there's always this question: can't centralized AI create the same advantage by hired, you know, data labelers, hiring this, hiring that? And our thesis on that has been always that, no. It cannot because there's too many different kinds of AI and data to be unlocked. And, as long as the first phase was all around data, which was also decentralized and locked by the Internet over decades. If there was no Internet decentralization, then there'll be no original model also.” - Himanshu (0:00)
  • “A lot of activities go into building a good model. So one has to collect data, filter data, then then then you train some smaller model and all that whole thing. That whole process, if you're actually building some useful AI, so Sentient has some useful AI it's targeting in the beginning, takes a lot of collaborative effort. As that collaborative effort is happening, you give every contributor ownership partial ownership of the model. And then you OMLize the model and release it through Sentient. These are Sentient models.” - Himanshu (0:00)
  • “So OML stands for open monetizable loyal, models and other artifacts. Okay. So that's that's what we started to build. That can be built as open models which can be openly shared, for a simple mod for for a simple mental picture, one can just think of models that can be downloaded and deployed. So they're open if you can download and deploy them. Right?... they have various definitions of open. Let's go with just this one. You download and deploy. K. So I'm free to finetune them. I'm free to use them that way. They are monetizable in the sense that when they are used, people who help build them, get the money. They get rewarded. Right? So that's what monetizable mean. And loyal is a very high bar. Loyal is someone who just you take it out of the environment, completely different place, but they never not do what they're not supposed to do. So they're always loyal to the builder, and even alignment of the model and safety rules of the model are all tied to what the builders prefer.” - Himanshu (0:00)
  • “So I'm a career academic. I've been in academia for a long time now. I was, I did my PhD in the US. I did my undergrad at, IIT Delhi, in Institute of Technology Delhi, and then I was in the US for about 8, 9 years. Then I was a faculty in industry of science in Bangalore, which is a top place in India. And around that time, I part so I my PhD is mostly on theory side and information theory and cryptography. But, and that's what I teach. But for the last decade or so, I've been building different systems, blockchain related and, and AI related. 4 years back, I reached out to my postdoc mentor, Pramod, that, I'm all tenured now, and I should do something more exciting.” - Himanshu (0:00)

🔒 More quotes available for premium users