Design for the AI we have,
not the one we envision.
the AI we have now is dictated by the paradigm of foundational LLMs,
they are
general-purpose
they lack
internal representation
they consume humongous
energy
they also
reinforce stereotypes
people don't seem to realize the limitations and issues of LLMs,
and I think it’s because we are designing the wrong interface
human interface finds a middle ground between what the user wants to do and what the technology can do
a good human interface informs the user
the limitation of the technology
best operational practices
potential means of trouble-shooting
the primary interface for LLMs is chatbot,
which achieves none of the above
instead, the interface of chatbot assumes that we can interact with LLMs as if we are interacting with humans
if LLMs can converse like humans, chatbot makes sense.
But the technology is just not there yet.
right now, human-AI interface is designed so that AI is a blockbox, making it impossible for the user to know what it is and how it works.
I want to design the embodiment of a different kind of AI that is more grounded to the affordance of it, so that we can have a proper conversation about the AI we want.