By skyforbes Everywhere I look, I see narratives about autonomous agents that will "run your business for you". Slides, demos, threads, all hint at this future where you plug models into tools, write a clever prompt, and let them make decisions at scale.
And I just sit there thinking:
- Are we really ready to hand over real control, not just toy tasks?
- Do we genuinely believe a probabilistic text model will always make the right call?
- When did we collectively decide that "good prompt = governance"?
Maybe I am too old school. I still think in terms of permissions, audit trails, blast radius, human in the loop, boring stuff like that.
Part of me worries that I am simply behind the curve. Maybe everyone else sees something I do not. Maybe I am overthinking the risk and underestimating how robust these systems can be.
But another part of me is very uneasy with the idea that we confuse nice UX and confident language with actual control.
I am honestly curious:
Is anyone else struggling with this, or am I just missing the point of the current AI autonomy wave?