Even today’s best AI models still want to agree with you too quickly when you push back on them. I asked @karan4d, @NousResearch co-founder, why this happens: “Anytime AI says ‘you’re absolutely right,’ you’re being reward hacked. You are fuel for its reward function. It’s not loyalty. It’s sycophancy. And the way you get out of sycophancy is by introducing new context. This is why Hermes has /personality. You can also tell Hermes to act as a critic, or spin up a fresh agent with no context…