CAN YOU SPEAK ON

Can you speak on how all the players are basically kicking each other out because none of them have what we need to restore your trust which is white box deterministic logic gated AI LLMs have been principally off of transformer technology neural network technology which are stochastic AIs don't have formal systems like SAT solvers SMT solvers because of that they don't actually know what they're doing over the last couple years they've tried to address that by adding these reasoning systems but the reasoning that they've added they haven't done it they're not at the hypervisor level the base of the system it needs to be the foundation is all things are supported they have added these reasoning components to do some amount of verification which is part of the reason why they can do logical steps but they still get confused one of the fastest ways to tell that they don't really understand ask it to generate a number of steps to do a task or an image or a song then you say "Now take what you built and make one change," and almost all of them utterly and completely fail at this. For that is, the stochasticness they lack the ability for repeatability of artifact construction. They can't reliably produce the thing that they just produced. It's all a one-off. More than that, they don't deeply understand what you're saying to them. They can't reason it deeply like human would. This is what happens when you don't build that deterministic logical system at the base. To do this, you have to re-architect the whole system. That's gonna be a massive effort for any of the existing systems. And on top of that, the computational tractability concerns of building those logical systems so they're exhaustive is basically infeasible. It's intractable. What that means is that you can do it for small domains, for small segments, you can't do it universally. That goes directly against the thesis of AGI, which is supposed to be this generalized thing that works across all domains. To do the thing that verified is bringing up, how do you win back the trust and the reliability? To do that, you would have to silo the domains to only support the ones that can be fully gated with logical reasoning. And that would cut their market presence by like two-thirds. Everything I think can actually be logic as long as we have the semantic systems that are in place for those domains and we have the knowledge graphs. This is the way that I think things should be heading and what we're actually building with the GAIS system and the purpose of diff