There's Nothing Artificial About It
I don't like the term artificial intelligence (AI). There is nothing artificial about it. LLMs derive vast amounts of real information. I call it Helpful Intelligence.
Helpful Intelligence
Helpful Intelligence aids me by automating routine work, analyzing vast amounts of data at high speeds, and improving decision-making across all aspects of my daily living and coding tasks.
For me, Helpful Intelligence will usher in a level of automation unseen before. The other huge benefit is smarter decisions. Using Helpful Intelligence I've been able to make smarter decisions, faster.
Software coding assistance is my most concrete example.
I don't use it blindly
I monitor AI very closely and carefully and I look for errors that only human judgement can spot. I don't use AI blindly.
I create doctrine first, that must be followed, and created a fail-safe catch, when change is happening at AI-speeds.
Doctrine first: privacy
The first and most important doctrine file was privacy. It underpinned all aspects of software architectural decisions I needed to make. Its first entry reads:
“The single source of truth for every KinectedCare decision involving user data. Read at every session start alongside CONTEXT.md, BESTPRACTICES.md, and PLAN.md.”
At times it was frustrating because of the constant checkins required. AI did not forget, nor did it allow me to deviate, without good reason. Over time my strict policy on Privacy was made better with the constant alignment during code development.
What AI does not have is human judgement. I was patient enough to come to a place where strict interpretation was easier, as my model to define privacy matured and was clear for a variety of use cases.
The doctrine is a circuit breaker
Privacy is also a good showcase of how software development, that was happening at AI speed, acted as a fail-safe catch. When software development occurs at AI speed, but there is doctrine that must be adhered to, the doctrine itself acts as a Human-in-the-Loop (HITL) circuit breaker. Instead of AI proceeding without intervention and human approval, it acts as a hard decision boundary.
Where my judgement made the call
The first circuit breaker was calling emergency authorities on behalf of the user. When and how to call emergency authorities first came at me at AI speed, with AI saying that it could not initiate that kind of call.
What happened next was a slow, precise, methodical and practical discourse with AI. I infused my judgement and intent.
I'd say it simply: if a person needed help and could not ask for it themselves, it was appropriate for the app to reach the people who care. When she presses HELP, or her watch reports a fall, KinectedCare alerts her CareTeam. She, or her watch, handles the emergency call.
Where it's heading
For me, AI is headed from simple chat to agentic workflows with multi-project, multi-purpose and multimodal.
Do better
Work Hard. Be Kind. Do Better. Helpful Intelligence lets me live those values at AI speed, and the doctrine keeps them from slipping. The technology is real, and so is my responsibility for how it's used. That's why I write the doctrine first, and why I'm the one who decides when it bends.