LOG30102022:Text to Image

Recently i had to generate a series of illustrations for a computer
graphics project. Fast iteration time was paramount, but so was
quality.
Rather than putting my art education to use and do the illustrations
myself, i decided to try some text to image models. Dall-E V2, and
later Midjourney, Stable Diffusion. We now reached a stage where for
some applications, the quality is good enough.

 From a commercial or economic perspective this was always going to
be implemented. Just as robotization was supposed to give us (or the
blue collar worker) more time to "enjoy life", "be with the family",
and so on. The same will be the case for this paradigm shift.
Except replace blue collar with mid class.

 Stock photo, image banks were already a vivid illustration of this
model. From the film days, custom assignments, to massification of
film photography in the late 80s, 90s. And the inevitable rise of
digital and greater massification with digital cameras. Now superceded
by mobile phone cameras.

 The amount of images produced is staggering, and for most
applications you don't need a 15 stop dynamic range, 50megapixel
image carefully calibrated with ColorChecker targets, store in ACES
AP0 or ACEScg AP1 AMPAS EXR IIF.

 Mobile suffices for most cases and image banks were scraping the
bottom of the barrel. Unless you have an army of slave labour
photographers working for peanuts, you're not going to make enough to
survive.

 Clearly some businesses saw this and decided to give the power to
mobile users and let them upload the images themselves and make a
pitance.
No names will be given, but you won't need to search for long to
find out who i'm referring to.

 This reference is made in order to illustrate the next step.
The continuation of the trend to replace creative industries by
automated systems that produce images with very fast iteration times,
in vast quantities, with "good enough quality".
Clearly illustrators are now the next stock photographers.
Not an extinct species yet, but clearly on their way to extinction.

Yet, there is something disturbing in this economic model. For
without the lower and middle class, who is going to support this
economic model?
If you now replace a group of illustrators by Stable Diffusion ML
models, or MidJourney, them what are they going to do?
These jobs aren't coming back.

A business i'm familiar with replaced a department of workers
responsible to calculate risk and do risk assessments in the
insurance industry, by a AI/ML model.
They're gone.

Actors are now scanning their faces so that they have good quality
digital doubles for the future. An immortal presence.

Yet, something is dystopian. Not for the technology itself, which
produces sometimes fantastically beautiful, even imaginative works of
art, if one can conceive that.
Clearly a blind test would give some of these AI/ML models top marks
in creativity, originality and execution.
The works are fabulous.

The issue is how unprepared we are to suddenly declare vast sectors
of economic production obsolete, and replace them by machines.
Machines that produce far more quantity, in faster iteration times.
All in all, this doesn't bode well for the future of the species.
Leaving aside urgent issues like climate change, or the
shortsightedness, stupidity and corruption of our politicians.

How wonderful it would be to have politicians of the caliber of an
Adenauer, De Gaulle, John F. Kennedy, even Kruschev.
We're lead by the blind and corrupt.

Perhaps we can replace the entire public sector by AI/ML models.
It can't possibly be worse than what we have right now.
In fact, i remember years ago reading a research paper that covered
the impact of random choices in political decision processes, versus
the choices of elected politicians. The paper was interesting because
it demonstrated that random choices were more positive for the
population on all time frames, than the choices made by elected
politicians.
We clearly need politicians with vision to survive rapid changes,
paradigm shifts even.
Right now, HAL9000 would be a better choice than we have.
Specially now. Amidst great change, uncertainty, turbulence and
apparently insurmountable challenges.