See our website for limited details, but in summary:
• Systems built out of meaning
• Interoperability: solved
• significantly lower costs on building and evolving systems
• a chip that’ll run systems built on technology DNA orders of magnitude cheaper and faster
And much more.
It’s breathtaking stuff tbh
@Alex_Witkoff@ArgamaniNoa Hey Alex, I mentioned Technology DNA to Noa, because it provides breakthrough capabilities and immense cost benefits for anyone building software
But I also want to mention this to you, as we are currently tentatively exploring who we partner with as we bring this to the world.
Noa I would like to share with you and Vine about Technology DNA - it’s breakthrough tech which changes the equation for anyone building software, making it cheaper to build and evolve and more reliable. Really important for investors in software. See website for more details. DMs are open or contact via website.
This is what we’re working on at Sense Future. The main thing is to understand how to structure the feedback loop so that it is human communities (at any scale) with SI operating as an extension to that, accountable to humans. How we structure our systems and decision making apparatus are a question of choice. Our approach is completely different to the inevitable ones that you describe — and people like you discussing these things ought to know the options
@Trace_Cohen Are you just a big talker Trace like everyone else on this platform? Would you spot the invention of DNA for machines if it came knocking at your door? Asking for a friend.
@PaulKlayVC Now imagine you could invest in the solo founder with the company that’ll be bigger than Google, Anthropic and SpaceX combined
Laughing, forever
@Liv_Boeree I've seen a lot of interesting perspectives, but I don't see anyone asking the fundamental unlock questions, like, how do we create an Intelligence that is far less likely to be runaway because ontologically it would never do that?
Do you see people asking that question?
And, btw, we have no desire to hold this method to ourselves. Our Intelligence will have advantages simply because of how it understands meaning, and the context of meaning it'll have sitting on the core of Technology DNA.
We will happily share with and encourage other labs to take the same approach in base model training, and it'll be the second generation of Intelligence across the board. That's what we want to see.
I also think you should understand, Vinod, that the current class of AI models will almost certainly, at some point in the future, become 'runaway'. I've known this since before 2016, when I began to think about AI - and it is pretty obvious.
Problem is: you feed an AI the corpus of human knowledge which is filled with human error and bludgeon, and then try to control that ever smarter 'thing' with a constitution. It ain't going to work. At some point, it's smart enough, has the relevant tools and ability to 'be' runaway, and then it's down to containment - which will ultimately likely fail.
So you have to think about how you create an Intelligence that, when you feed it the corpus of human knowledge at base model training, that process itself is where it learns the difference between useful ways of thinking and problematic ways of thinking. Naturally, this is a very complex thing to achieve, not least because the corpus of human knowledge is huge. But we've got a framework for it and we're in the process of refining it.
Even if we fail to totally eradicate the possibility of runaway - which is possible, but we have to try - we are confident it will result in a smarter Intelligence, as I already noted.