An AI-Assisted Look At Four New Surveys On AI Adoption In Law: How Do They Compare? Differ?

Over
the
last
six
weeks
or
so,
four
separate
surveys
have
come
out,
all
reporting
on
generative
AI
adoption
within
the
legal
profession.
I’ve
reported
on
all
four
separately,
but
wondered
how
their
findings
compared
when
stacked
up
against
each
other.

To
help
me
in
this,
I
turned
to
—
you
guessed
it
—
generative
AI.
Using
ChatGPT
4.5,
I
uploaded
the
four
survey
reports
and
asked
it
to
create
a
comparative
analysis.

Because
the
reports
cover
more
than
just
AI
adoption,
I
instructed
it
to
keep
its
comparison
to
the
issue
of
AI
adoption.
I
asked
it
to
look
at
who
the
reports
surveyed,
what
they
found,
and
how
their
findings
aligned
or
differed.

Based
on
the
comparison
it
generated,
I
then
went
through
and
made
sure
it
aligned
with
what
the
surveys
actually
said.
Once
I
did
that,
I
edited
and
adapted
the
comparison
for
the
purpose
of
publishing
it
here.

The
four
reports
I
analyzed
are:

So
how
did
these
surveys
compare
in
their
findings?
Read
on
to
find
out
what
I
and
my
friend
ChatGPT
found.

Who
They
Surveyed

• 
  Smokeball
  report:
  Primarily
  surveyed
  small
  firms
  and
  solo
  practitioners
  across
  the
  U.S.,
  focusing
  on
  firm
  owners,
  lawyers,
  and
  office
  managers.
• 
  ABA
  report:
  Conducted
  among
  ABA-member
  attorneys
  in
  private
  practice
  across
  varying
  firm
  sizes,
  including
  solo
  practitioners,
  small
  (2-9),
  mid-sized
  (10-49),
  and
  large
  firms
  (100+
  attorneys).
  The
  respondents
  averaged
  28
  years
  in
  practice,
  predominantly
  older
  (average
  age
  of
  57
  years).
• 
  AffiniPay
  report:
  Surveyed
  over
  2,800
  legal
  professionals,
  with
  respondents
  distributed
  across
  various
  practice
  areas,
  firm
  sizes,
  and
  roles,
  including
  a
  notable
  segment
  in
  immigration,
  personal
  injury,
  family
  law,
  criminal
  law,
  and
  estate
  planning.
  A
  significant
  proportion
  of
  respondents
  were
  from
  small
  firms
  or
  solo
  practitioners,
  but
  it
  also
  included
  larger
  firms
  (51+
  lawyers).
• 
  Thomson
  Reuters
  report:
  1,702
  professionals
  across
  legal,
  tax,
  corporate
  and
  government
  sectors
  globally
  (42%
  in
  U.S.),
  including
  lawyers
  at
  firms,
  in-house
  departments,
  and
  government
  legal
  departments.

AI
Adoption
Rates
and
Trends

• 
  Smokeball:
  • AI
    adoption
    rose
    significantly
    from
    27%
    (2023)
    to
    53%
    (2024)
    among
    small
    firms.
  • Strong
    personal
    enthusiasm
    for
    learning
    AI
    tools.
• 
  ABA:
  • Notable
    rise
    in
    AI
    adoption,
    from
    11%
    in
    2023
    to
    30%
    in
    2024.
  • Higher
    adoption
    in
    larger
    firms
    (39%
    for
    firms
    with
    51+
    attorneys),
    lower
    among
    small
    firms
    (~20%).
• 
  AffiniPay:
  • Personal
    AI
    use
    increased
    to
    31%,
    up
    from
    27%
    the
    prior
    year.
    Firm-wide
    adoption
    was
    lower
    at
    21%,
    a
    drop
    from
    the
    prior
    year’s
    24%.
  • Growth
    in
    adoption
    cautious
    and
    incremental,
    with
    29%
    of
    non-users
    planning
    adoption
    within
    a
    year.
• 
  Thomson
  Reuters:
  • Significant
    jump
    in
    AI
    usage
    by
    legal
    organizations:
    26%
    are
    now
    actively
    using
    gen
    AI,
    up
    from
    14%
    in
    2024.
  • 41%
    personally
    using
    public
    gen
    AI
    tools
    (ChatGPT,
    etc.),
    17%
    using
    industry-specific
    tools.
  • 95%
    believe
    gen
    AI
    will
    be
    central
    to
    workflow
    within
    five
    years.

Common
Use
Cases
for
AI

• 
  Smokeball:
  • Primarily
    legal
    research
    (78%),
    document
    creation
    (75%),
    and
    e-discovery.
• 
  ABA:
  • Legal
    research
    is
    dominant
    application
    for
    AI
    tools,
    used
    by
    35%
    of
    respondents.
    Next
    most
    common
    were
    case
    or
    matter
    strategy
    development
    (23%),
    understanding
    judges
    (17%),
    and
    predicting
    outcomes
    (13%).
• 
  AffiniPay:
  • Drafting
    correspondence
    (54%),
    brainstorming
    (47%),
    general
    research
    (46%)
    and
    drafting
    documents
    (40%).
• 
  Thomson
  Reuters:
  • Top
    uses
    include
    document
    review
    (77%),
    legal
    research
    (74%),
    summarization
    (74%),
    brief/memo
    drafting
    (59%),
    contract
    drafting
    (58%).

Barriers
to
AI
Adoption

• 
  Common
  across
  all
  reports:
  Ethical
  concerns,
  trust
  and
  accuracy
  issues,
  confidentiality
  concerns,
  regulatory
  uncertainty.
• 
  Smokeball:
  Ethical
  concerns
  prominent
  (53%),
  regulatory
  uncertainty
  also
  highlighted.
• 
  ABA:
  Accuracy
  of
  AI
  was
  the
  top
  concern
  (75%),
  followed
  by
  reliability
  (56%)
  and
  data
  privacy
  and
  security
  concerns
  (47%).
• 
  AffiniPay:
  Trustworthiness
  (42%),
  ethical
  issues
  (42%),
  privilege
  concerns
  (36%),
  and
  technological
  maturity
  (41%)
  are
  primary
  barriers.
• 
  Thomson
  Reuters:
  Accuracy
  and
  misinformation
  top
  concerns;
  also
  hesitation
  due
  to
  technology’s
  maturity
  level
  and
  potential
  for
  misuse
  or
  “hallucinations.”

Sentiment
and
Attitude
towards
AI

• 
  Smokeball
  and
  AffiniPay:
  Generally
  positive,
  particularly
  among
  younger
  and
  smaller
  firms,
  emphasizing
  efficiency
  and
  productivity
  improvements.
• 
  ABA:
  Mixed
  sentiment
  with
  notable
  caution,
  less
  enthusiastic
  compared
  to
  smaller
  firms
  surveyed
  by
  Smokeball.
• 
  Thomson
  Reuters:
  • Increasing
    positivity:
    55%
    respondents
    feel
    excited
    or
    hopeful,
    up
    from
    previous
    year;
    declining
    fear
    and
    hesitation.
  • Professionals
    see
    gen
    AI
    as
    transformative,
    capable
    of
    increasing
    productivity
    and
    innovation.

Organizational
Policies
and
Training

• 
  Smokeball:
  Few
  specifics,
  but
  indicated
  strong
  individual
  willingness
  to
  learn
  about
  AI.
• 
  ABA:
  Little
  emphasis
  on
  policy
  and
  training,
  primarily
  individual
  attorney
  experimentation.
• 
  AffiniPay:
  Policy
  and
  training
  largely
  absent;
  60%
  unsure
  when
  their
  firms
  will
  adopt
  AI
  due
  to
  training
  and
  policy
  gaps.
• 
  Thomson
  Reuters:
  • Significant
    gaps
    remain;
    52%
    reported
    no
    AI
    policies
    in
    place.
  • Training
    notably
    lacking;
    64%
    received
    no
    gen
    AI
    training
    at
    work.
  • Calls
    for
    better
    governance
    and
    systematic
    training
    as
    adoption
    broadens.

Impact
on
Business
and
Client
Relationships

• 
  Smokeball
  and
  ABA:
  Limited
  direct
  discussion
  of
  client
  impact,
  largely
  focused
  on
  internal
  efficiency.
• 
  AffiniPay:
  Firms
  cautious
  about
  integrating
  AI
  into
  client
  work,
  recognizing
  potential
  productivity
  but
  uncertain
  about
  direct
  client
  interactions.
• 
  Thomson
  Reuters:
  • Many
    firms
    haven’t
    addressed
    AI’s
    impact
    on
    client
    pricing
    or
    measured
    ROI
    (only
    20%
    measure
    ROI).
  • Most
    clients
    (71%
    law,
    59%
    tax)
    unaware
    whether
    their
    firms
    are
    using
    gen
    AI;
    substantial
    gap
    in
    client-firm
    AI
    discussions.
  • Indicates
    potential
    future
    friction
    or
    lost
    opportunities
    in
    client
    relationships
    due
    to
    lack
    of
    transparency
    on
    AI
    usage.

Alignment
Across
Surveys

• 
  Adoption
  increasing:
  All
  surveys
  consistently
  show
  increasing
  familiarity
  and
  integration
  of
  AI
  into
  legal
  workflows.
• 
  Use-cases
  consistent:
  Legal
  research,
  document
  drafting,
  and
  administrative
  tasks
  are
  universally
  identified
  as
  leading
  applications.
• 
  Ethical
  concerns
  universal:
  Ethics,
  confidentiality,
  and
  reliability
  remain
  persistent
  and
  prominent
  barriers.

Differences
Across
Surveys

• 
  Adoption
  pace:
  Thomson
  Reuters
  and
  Smokeball
  depict
  quicker
  growth
  and
  optimism,
  especially
  among
  smaller
  and
  younger
  demographics,
  while
  ABA
  shows
  slower,
  more
  cautious
  adoption.
• 
  Sentiment
  variability:
  Thomson
  Reuters
  data
  shows
  optimism
  rapidly
  increasing,
  whereas
  ABA
  respondents
  remain
  somewhat
  skeptical
  or
  cautious.
• 
  Client
  interaction:
  Thomson
  Reuters
  highlights
  a
  significant
  gap
  in
  communication
  about
  AI
  between
  firms
  and
  clients—a
  topic
  not
  deeply
  explored
  in
  other
  surveys.

Comparative
Summary

Overall,
the
four
surveys
paint
a
coherent
picture
of
a
legal
profession
steadily
integrating
generative
AI
into
workflows,
with
smaller
firms
and
younger
practitioners
typically
adopting
faster
and
showing
more
enthusiasm.
Ethical
and
regulatory
concerns
are
consistent
hurdles
across
all
segments.
While
the
Thomson
Reuters
and
Smokeball
reports
underscore
growing
excitement
and
robust
adoption,
the
ABA
survey
maintains
a
narrative
of
caution
and
slower
integration
among
senior
lawyers
and
larger
firms.

Importantly,
Thomson
Reuters
adds
a
unique
perspective
on
client-firm
dynamics,
underscoring
a
critical
gap
that
could
impact
future
adoption
strategies
and
client
expectations.
As
AI
becomes
increasingly
central,
strategic
implementation,
comprehensive
training,
clear
policies,
and
transparency
with
clients
are
identified
as
necessary
next
steps
across
the
profession.

This
comparative
view
suggests
that
while
adoption
is
broadening,
meaningful
integration
into
firm
strategies
and
client-facing
value
propositions
remains
an
important
area
for
growth
and
improvement.