This gets way more critical with brain interfaces. A stray thought could trigger unintended actions. But just like sports or music, we can probably train ourselves to reduce noise and improve precision over time
Speech is a massively high-bandwidth interface. Ever compare how fast you talk vs. type? Now imagine thumbs vs keyboard vs speech vs brain. What’s the true upper bound of bandwidth in Human-Computer Interaction (HCI)?
An old prof once told me that In #HCI, a low false positive rate can be really tricky to get, and having too high a false positive rate can make the interface experience horrible.
Been having a real fun time working with #Gemini to refine requirements and gather context for .md files that are then ingested by #claudecode to implement :)
#AI may raise us to a consistently higher, less rigid level of abstraction to where it feels awkward to describe what we are doing as programming, but #systemdesign will always remain.
Ideally the more powerful the AI, the less specific we have to be to get a useful output, and the more efficient the iteration process towards a truly valuable result (and hopefully more fun 🥲)
"Denotational design with type class morphisms" by #ConalElliot is one of my favourite design related reads. In my experience developing systems with the lens it describes has been fun, less stressful, and more value producing, especially when paired with a focus on iterating
"Denotational design with type class morphisms" by #ConalElliot is one of my favourite design related reads. In my experience developing systems with the lens it describes has been fun, less stressful, and more value producing, especially when paired with a focus on iterating
I don't know how discrete the separation will turn out in practice though, the AIMs feel much more of a safe interface until we understand more about #consciousness and its role in long term progress and horizon-like complexity and goals/influence
The differences between the AI Bobs (human consciousness, emotional) in the We Are Legion (We Are Bob) books and the AIMs (functional and administrative) provides an interesting frame for AI progression. Both are insanely useful, but in very very different ways #wearelegion#AI
@AdamRackis spreading seems to convert tuple into an array. Probably something due to the specific semantics of the spread operator and how typescript may not be enable to enforce array positioning / length with its application? Im not sure...