What
do
you
do
when
your
client
knows
more
about
the
law
and
strategy
than
you
do?
Or
at
least
thinks
they
do.
That’s
the
dilemma
a
lot
of
lawyers
and
legal
professionals
face
these
days.
Clients
come
into
the
office
armed
with
all
sorts
of
advice
and
information
they
have
obtained
from
publicly
facing
AI
systems,
information
which
is
often
just
wrong.
Let’s
Face
It,
Clients
Are
Using
AI
I
was
talking
to
a
couple
of
criminal
lawyers
recently
about
this
very
subject.
They
say
it
happens
all
the
time.
And
it’s
not
just
that
the
clients
have
information
that
may
or
may
not
be
correct
or
even
helpful.
It’s
the
confidence
that
the
clients
seem
to
have
in
that
information.
Which
makes
the
lawyer’s
job
that
much
harder
when
the
information
or
recommended
strategy
is
wrong.
It
takes
time
away
from
the
case
handling
to
constantly
try
to
convince
the
client
that
the
bot
is
off
base.
A
bot
that,
of
course,
says
things
with
such
confidence.
I’m
sure
it’s
the
same
on
the
civil
side.
The
problem
is
not
just
one
of
convincing
the
client
that
what
they
got
from
the
bot
is
bad,
it’s
the
seeds
of
mistrust
that
are
being
sown.
The
law
and
strategy
are
often
gray
areas.
Many
times,
there
is
no
clearly
right
answer,
no
foolproof
strategy
or
guaranteed
outcome.
The
best
answers
often
stem
from
the
lawyers’
combined
experience,
judgment
and
know
how,
much
of
which
can’t
be
duplicated
by
a
bot.
If
I
Had
A
Better
Lawyer
But
when
the
client
begins
to
wonder
that
the
lawyer’s
recommendations
are
wrong
based
on
what
a
bot
tells
them,
the
client
starts
to
think:
I
would
be
getting
a
better
result
if
I
had
a
better
lawyer.
A
lawyer
who
would
do
what
ChatGPT
recommends.
Add
to
this
the
fact
that
all
too
often
AI
tools
will
tell
a
person
what
it
thinks
the
person
wants
to
hear.
And
you
have
real
and
nagging
problem.
It’s
a
problem
that
will
lead
to
a
lot
of
time
and
energy
being
spent
trying
to
convince
the
client
that
they
should
be
listeing
to
you,
not
ChatGPT.
And
it’s
a
problem
not
just
at
the
beginning
of
the
case
when
the
client
walks
in
the
door.
It
runs
throughout
the
case.
If
the
client
goes
to
ChatGPT
before
they
see
their
lawyer,
you
can
bet
that
they
are
going
to
go
back
to
ChatGPT
every
time
the
lawyer
tells
them
something
they
don’t
want
to
hear
or
don’t
believe.
Which
will
lead
to
more
and
more
arguments,
distrust,
and
disruption
for
the
human
lawyer.
Not
to
mention
the
fact
that
the
client
may
be
revealing
attorney-client
privileged
information
in
its
discussions
with
the
bot.
As
I
have
previously
discussed,
placing
confidential
material
in
a
public
AI
platform
risks
waiving
the
privilege
since
it
may
no
longer
be
considered
confidential.
It’s
hard
enough
for
lawyers
to
use
AI
and
avoid
this
waiver.
It
may
be
well-nigh
impossible
for
a
client
trying
to
get
the
answer
they
want
to
hear
to
not
tell
the
bot
more
than
they
should.
Using
AI
To
Turn
The
Tables
One
of
the
criminal
lawyers
I
was
talking
to
did
have
a
bit
of
a
solution,
however.
His
strategy
when
a
client
tells
him
something
that
came
from
a
bot
that’s
wrong
is
to
ask
the
client
about
the
input
they
provided.
He
then
tries
to
show
them
that
with
a
proper
input,
asking
the
right
question,
a
more
accurate
result
can
be
obtained
that’s
often,
he
says,
in
line
with
his
advice.
He
also
shows
the
client
how
by
working
with
the
prompt
to
suggest
a
desired
result
affects
the
outcome.
Of
course,
successfully
doing
what
this
lawyer
suggests
takes
time,
work
and
effort.
It
also
requires
a
good
understanding
of
how
AI
works
and
how
to
construct
prompts.
But
it’s
better
than
having
an
ongoing
battle
with
the
client
over
AI
outputs.
And,
frankly,
it
also
takes
some
courage
since
there’s
a
risk
that
the
bot
may
be
right
and
the
lawyer
wrong.
That’s
a
reality
that
is
uncomfortable.
But
if
so,
so
much
the
better.
By
constructing
good
prompts,
AI
sometimes
comes
up
with
good
stuff
and
we
owe
it
to
our
clients
to
be
open
to
different
views.
Frankly,
the
lawyer
should
be
checking
their
views
and
ideas
against
AI
tools
on
the
front
end
to
be
prepared
for
the
client
armed
with
AI-generated
information.
I
like
the
idea.
It
saves
time.
It
reduces
disruption.
It
breeds
trust.
I
found
over
the
years,
for
example,
that
if
you
can
anticipate
what
a
client
thinks
and
will
say
and
then
show
them
why
they
are
misguided,
it
enhances
their
confidence
in
you. And
using
AI
in
an
educational
way
moves
the
relationship
more
to
one
of
a
team
with
a
lawyer,
client,
and
AI
tools
all
working
together
to
get
better
results
with
less
friction.
And
by
the
way,
it
squares
with
our
ethical
duty
under
the
competency
Rule
1.1
to
know
and
understand
the
benefits
and
risks
of
technology.
That
applies
not
only
to
the
technology
we
use
as
lawyers
but
the
technology
we
can
reasonably
expect
our
clients
will
be
using.
And
today,
that’s
AI.
Our
clients,
like
virtually
everyone
else,
are
going
to
be
consulting
AI
about
their
legal
problems
and
what
we
are
telling
them.
So
we
better
be
using
AI
as
well.
Not
using
tools
our
clients
are
using
can
lead
to
a
very
tense
and
unsatisfying
attorney
client
relationship.
Practicing
law
is
hard
enough,
let’s
not
make
it
harder.
So
let’s
get
real
and
meet
our
clients
where
they
are.
When
the
client
walks
into
your
office,
be
ready
with
your
bot.
Stephen
Embry
is
a
lawyer,
speaker,
blogger,
and
writer.
He
publishes TechLaw
Crossroads,
a
blog
devoted
to
the
examination
of
the
tension
between
technology,
the
law,
and
the
practice
of
law.