As AI agents become more autonomous, the key challenge isn't what they can do; it's how to design the human side of the equation.
From #MeaningAwareAI to #QuantumMinds & Beyond... Learn from #HumanCenteredMeaningAwareAI Pioneer:https://www.linkedin.com/pulse/us-federal-reserve-treasury-goldman-sachs-jp-morgan-global-malhotra-8gvre/ starting with the very first paper on the topic: Malhotra, Y., Expert Systems for Knowledge Management: Crossing the Chasm between Information Processing and Sense Making, Expert Systems with Applications: An International Journal, 20(1), 7-16, 2001. https://www.brint.com/expertsystems.pdf : Latest on #MeaningAwareAI #QuantumMinds: https://www.youtube.com/playlist?list=PLXz9OqWahsHowKYoKuNGnwNLX_COKrhQ4
Ozzy Aghedo
2w
Also need to find a way to broker access so the agents don't get hold of the raw credentials... who knows what it could do outside its scoped environments
Excited to see this highlighted. Thanks for sharing !! Cc James Pierce Vaiva Kalnikaitė
Amazon Science: Important point. But here’s what must be made explicit: Outcome liability. Because “designing the human side” isn’t just about: – usability – trust – interaction It’s about: 👉 Who owns the outcome when autonomous agents act. As agents become more autonomous: – they make decisions – they trigger actions – they produce real-world consequences And the “human side” becomes: the assignment of responsibility. Right now, the focus is on: ✔ human-AI interaction ✔ oversight ✔ alignment But missing: ✖ who carries outcome liability when agents act ✖ how responsibility is defined before execution ✖ what happens when the system is wrong Because autonomy shifts the burden. Not away from humans. Onto them. So the real design question isn’t just: “How do humans interact with agents?” It’s this: 👉 Where does outcome liability sit when the agent makes a decision? If that’s unclear: you’re not designing the human side. You’re distributing action without ownership.