About this site
Meeting New Challenges Head-On
When approaching how to update this site, reinventing your portfolio after fifteen official years
of one's career calls for something...more.
I’m finely attuned to the fact that the tech world is
currently obsessed with AI while the world looks on
in silent horror of this new frontier of
industrialization. Personally, I've found ways to leverage
it in ways that vastly extend the reach of my
capabilities-- for personal projects.
I've learned since that ChatGPT is pretty good at coding,
even if it makes mistakes and has the memory of a goldfish
sometimes.
If the generative AI can get me 90% of the way, is it
not worth figuring out the last 10%?
I come at AI from the point of view of someone who has
always been an aspiring illustrator for far longer than
a Product Designer.
I value painstakingly painted canvases with tens of
thousands of hours of experience making the art, personal
expression over facsimile images generated from a
database by a prompt that then translates into the
lookup table to generate an approximation of what your
prompt described.
But that doesn't mean it cannot be very useful
for a variety of use cases where traditional art
wouldn't have been employed previously anyway.
Without traditional art, there's nothing to mimic.
All day everyday now, for far more than
the last decade and with increasing frequency, we are all glued to glowing screens that affect our
circadian rhythm in unnatural ways.
We are human, and must pay heed not to
squander it.
On Generative AI
My first real foray with Generative AI was in the
pre-production for my short film, Second Land.
I knew what I was looking for, but by myself I can only
do so much, and pre-production art isn’t something to be
plastered on a wall, it’s to grow ideas quicker.
I used Midjourney to get an image that I liked,
that I then edited extensively in photoshop to more
closely match what I had in mind. Used in a collage
manner, it enables your ideas to have wider reach in a
shorter amount of time, which then enables all sorts of
other advantages and opportunities.
So when I first imagined what I would find engaging
for my portfolio, I thought about AI Chatbots.
I did not want to integrate an existing service
as it wouldn’t be of much credit to me as a
Product Designer.
If I am to leave a memorable impression, I would have to
find a way to put everything I’ve learned on display at once;
from basic front-end skills, to hinting at my
illustration, to developing the entire experience as
a cohesive product, or coming as close as I can with
the time and resources available to me.
Inspiration
Like many others of my generation I grew up with a
portion of my sense of humor shaped by late 80s/early 90s
LucasFilm/LucasArts adventure games, which were always notable
for doing a lot with very little, both in terms of technical
limitations as well as in economy.
A 8mb game could provide many hours of engaging
interactive fun.
I’ve also always liked how in RPG games the developers
augment speech with text being typed out.
I would leverage ChatGPT for the code to enable my vision.
My site has been built on top of SquareSpace for a very
long time, but my was put in "developer mode", so it's
benefiting from an outdated feature set.
Sometimes there's just no time to rebuild from scratch.
Having been involved with HTML/CSS since maybe 9 or 10
years old helped me contextualize the code being
generated so I could add it and modify it for my usecase.
As anyone who's used it in any serious extent knows,
ChatGPT is prone to errors in context handling roughly
proportionately to increasing complexity.
Just as with Stable Diffusion based Generative AI, how the prompt is
composed, in what order and what emphasis where determines
the outcome of the code being generated. Predictably the
more complex the code, the harder it is for ChatGPT to
handle, regardless of the context.
To understand why
this is, one can look at a Stable Diffusion prompt being
described with too many imposed keywords with weights that
the model struggles to string together in any coherent way.
Too many things to try to reconcile together,
perhaps because they are or are interpreted as
incompatible and the output becomes its best guess.
It took me roughly 8 weeks of countless iterations until I got a
script that met all my criteria. As with any and all
software development, sometimes there are forks in the
road, you either get feature A or feature B but together
they may not work for
.
I’ve devised that the helper
tool on the root of the resumé serves a
than a helper tool on a page, as during testing I found
that this tool can quickly become incredibly annoying.
That’s why pages have a different version of the tool that is
specialized and minimized to not be obnoxious but to
provide any context if a reader should hover over
something.
I've only added the helper to the root, this page and the
cisco writeup, as everything else is too archaic to
make accurate statements about a decade later.
Whether or not my silly little tool left you with a
lasting impression, I know that I would not have been
able to cram as many features into it as I have, and
would no doubt have to make many sacrifices in my
to complete it.
Generative AI is a tool. If you know how to use it,
it’s an incredible enablement tool that creates lots of
opportunities and advantages. It doesn't have rational
decision making abilities that adapt to circumstances.
That's the human niche.
I hope in time, AI will be regarded as a solution
for scalability rather than as means to replace
humans. Alas even now, humans can use AI for their
enormous benefit. It helps to have extracurricular
interest in it .
I will always trust flesh and blood engineers over
generated code, my understanding of the current
sentiment on AI in the coding sphere seems to be
mostly around debugging the generated code code,
and indeed, I’ve spent majority of the development
time debugging the steps back after the steps forward.
It’s been a good two months of heads down getting
everything sorted, but it's certainly felt worth the
effort.
As a designer, I now have the tools to prototype out
anything I can imagine, based on my very own organic
database of experiences and decades of broader perspectives.
AI has removed any and all roadblocks from my vision.