Remember
the
first
time
you
opened
a
copy
of
the
Federal
Register
to
feast
upon
eight
pages
of
tiny
print
explaining
the
proper
size
of
a
gasket?
Did
you
ever
think,
what
if
they
replaced
all
the
highly
trained
engineers
and
scientists
describing
the
precision
crafting
required
to
keep
airplanes
from
falling
out
of
the
sky
with
hallucinating
robots?
Well,
good
news!
ProPublica dropped
the
investigative
piece this
morning,
revealing
that
the
Trump
administration
plans
to
use
Google
Gemini
to
draft
federal
transportation
regulations.
Turning
over
to
a
chatbot
the
very
nuts
and
bolts
of
the
agency
responsible
for
keeping
the
nation’s
highways
safe
may
scare
some
people,
but
really,
who
can
be
better
trusted
to
understand
the
unique
challenge
of
transportation
rulemaking
than
Road
Rules/Real
World
Challenge
champion
Sean
Duffy?
Road?
Rules?
Challenge?
It’s
all
right
there
on
his
resume!
This
isn’t
necessarily
a
bad
idea.
Assuming
the
underlying
expertise
and
science
remains
sound,
AI
tools
actually
excel
at
the
job
of
converting
technical
information
into
approachable
and
readily
comprehensible
text.
Rulemaking
has
always
involved
injecting
relative
clarity
into
complex
information,
a
mission
that’s
even
more
important
in
light
of
a
Loper
Bright
world.
Without
judges
deferring
to
agencies,
there’s
even
more
urgency
to
eradicate
ambiguities
and
deliver
rules
that
even
the
most
out-of-their-depth
judges
could
understand.
In
a
perfect
world,
regulatory
lawyers
would
use
AI
to
draft
clear
pronouncements
that
explicitly
spell
out
contingencies
that
were
once
left
to
deference.
As
a
preliminary
drafting
tool,
married
to
the
legal
acumen
of
veteran
practitioners,
AI
could
make
rulemaking
better
for
all.
It
will
shock
you
not
at
all
to
learn
that
this
is
not
the
Trump
administration’s
logic.
In
an
internal
meeting
announcing
the
new
initiative,
ProPublica
reports
that
DOT
General
Counsel
Gregory
Zerzan
expressed
enthusiasm
for
the
algorithmic
outsourcing,
noting
“We
don’t
need
the
perfect
rule
on
XYZ.
We
don’t
even
need
a
very
good
rule
on
XYZ.
We
want
good
enough.”
Do
you
feel
safer
already?
“We’re
flooding
the
zone,”
Zerzan
continued,
invoking
Steve
Bannon’s
classic
advice
for
the
MAGA
movement:
keep
doing
more
and
more
outrageous
nonsense
to
prevent
anyone
from
having
time
to
push
back.
It’s
the
same
mentality
that
might
kidnap
a
foreign
leader,
send
shock
troops
to
murder
civilians
in
Minnesota,
and
attempt
to
take
over
Greenland
so
everyone
will
forget
that
the
Epstein
files
haven’t
been
released
as
required
by
statute.
Zerzan’s
implication
is
that
the
Trump
Department
of
Transportation
wants
to
gut
federal
transportation
safety
regulations
and
they
want
to
do
it
by
bombarding
the
public
with
so
many
changes
that
it
doesn’t
have
time
to
ask
why.
If
it
results
in
some
hallucination-filled
half-assery,
the
administration
is
willing
to
live
with
that.
Mike
Horton,
DOT’s
former
acting
chief
artificial
intelligence
officer,
criticized
the
plan
to
use
Gemini
to
write
regulations,
comparing
it
to
“having
a
high
school
intern
that’s
doing
your
rulemaking.”
(He
said
the
plan
was
not
in
the
works
when
he
left
the
agency
in
August.)
Noting
the
life-or-death
stakes
of
transportation
safety
regulations,
Horton
said
the
agency’s
leaders
“want
to
go
fast
and
break
things,
but
going
fast
and
breaking
things
means
people
are
going
to
get
hurt.”
I’m
sorry,
you’re
absolutely
right!
We
needed
better
brakes
to
keep
the
train
from
careening
off
the
track
into
that
schoolhouse
and
dumping
toxic
sludge
into
that
reservoir.
Do
you
want
me
to
generate
another
rule
based
on
this
new
information?
But
can
the
administration
really
use
AI
to
deliver
the
Mad
Max-inspired
highway
hellscape
they
seek?
I’m
skeptical.
This
administration
has
already
run
afoul
of
AI
understanding
the
rule
of
law
better
than
its
cynical
Federalist
Society
approved
handlers.
The
Pentagon
installed
a
dedicated
AI
and
it
instantly
identified
war
crimes.
Arguably,
the
only
thing
more
difficult
than
keeping
AI
from
making
mistakes
is
convincing
it
to
intentionally
produce
garbage.
The
thing
about
a
“high
school
intern,”
as
Horton
put
it,
is
that
it
can
struggle
to
act
in
bad
faith
because
it
doesn’t
know
enough
to
simulate
guile.
Lawyers
may
find
themselves
substantially
rewriting
everything
the
AI
puts
out
anyway.
And
even
if
they
do
rapidly
churn
out
rulemaking
copy,
the
protections
found
in
the
Administrative
Procedures
Act
still
exist.
Between
notice
and
comment
and
judicial
review,
the
public
will
still
have
mechanisms
to
put
the
brakes
on
the
—
in
this
case
literal
—
runaway
train.
Zerzan
wants
to
flood
the
zone
but
might
only
succeed
in
spreading
himself
out.
These
rules
will
have
different
constituencies,
but
the
Department
will
be
involved
in
every
fight.
It
doesn’t
“flood
the
zone”
to
have
the
shipping
industry,
the
trucking
industry,
and
airlines
all
involved
in
fights
during
the
same
month.
From
their
perspective,
they’re
all
just
having
one
fight.
But
the
Department
is
defending
three
fights
at
once.
And,
of
course,
when
you
“go
fast
and
break
things,”
it
usually
ends
in
a
litigation
bottleneck.
The
Department
of
Transportation
already
tried
its
hand
at
rulemaking
the
old-fashioned
way.
Pumping
out
a
rule
in
six
months
in
an
effort
to
strip
licenses
from
a
couple
hundred
thousand
truck
drivers
as
part
of
the
administration’s
ongoing
nativism
performance
art,
the
Department
only
succeeded
in
getting
it
blocked
by
the
D.C.
Circuit.
This
effort
wasn’t
going
to
be
any
more
successful
if
they’d
pushed
the
rule
out
the
door
faster.
If
anything,
flooding
the
zone
will
just
clog
up
the
courts
trying
to
untangle
the
lawsuits.
Also…
have
fun
with
the
discovery
requests
seeking
every
prompt
lawyers
used
in
the
drafting
process.
According
to
ProPublica’s
sources,
as
part
of
downplaying
the
serious
repercussions
of
the
proposal,
staffers
were
told
that
a
lot
of
what
the
AI
would
be
producing
would
be
“word
salad.”
Hard
to
assert
any
sort
of
privilege
over
“word
salad.”
All
this
“flood
the
zone”
talk
is
just
posturing.
They
want
to
front
like
they’re
launching
some
nefarious
strategic
blitzkrieg
on
the
administrative
state,
but
they
just
want
AI
to
write
these
regs
for
them
because
they’re
fucking
lazy.
Remember
when
Elon
Musk
parachuted
into
the
government
promising
to
cut
a
trillion
dollars
in
spending,
destroyed
some
agencies,
definitely
killed
hundreds
of
thousands
of
people,
and
ultimately
walked
away
having
left
the
deficit
worse
than
before?
It’s
the
same
energy.
Lawyers
working
in
tandem
with
AI
technology
can
marginally
enhance
their
productivity,
but
consider
the
rhetoric
they’re
using
according
to
this
report:
“flooding
the
zone,”
“point
of
the
spear,”
“shouldn’t
take
you
more
than
20
minutes.”
Too
often
we
construe
their
contempt
for
the
work
of
government
as
a
problem
with
the
work
government
performs.
But
a
lot
of
the
animosity
is
based
on
a
contempt
of
the
idea
of
work
itself.
Of
the
mundane,
nose-to-grindstone,
detail-oriented
work
that
governments
—
and
lawyers
—
have
to
do.
Which,
in
the
end,
is
AI’s
greatest
threat
to
society.
It
enthralls
the
nation’s
laziest.
Government
by
AI?
Trump
Administration
Plans
to
Write
Regulations
Using
Artificial
Intelligence [ProPublica]