What you comment on around software guarantees is key here. When you chain together AI systems where bad outputs become bad inputs failures cascade and blow up. The lack of guarantees in AI is a feature, not a bug which limits its usefulness where guarantees (such as a system for provisioning electricity across a grid) are needed.
Yep I want to expand on the point around software guarantees in a future post. I think it's critical for making any ML-powered system work in practice.
What you comment on around software guarantees is key here. When you chain together AI systems where bad outputs become bad inputs failures cascade and blow up. The lack of guarantees in AI is a feature, not a bug which limits its usefulness where guarantees (such as a system for provisioning electricity across a grid) are needed.
Yep I want to expand on the point around software guarantees in a future post. I think it's critical for making any ML-powered system work in practice.