The Intermediary – October 2026 - Flipbook - Page 84
T E C H N O L O GY
Opinion
We must nurture the
future of judgement
W
e have spent a
great deal of
time discussing
what artificial
intelligence
(AI) might
take away from people working in
financial services. Perhaps the more
interesting question, though, is what
it gives back. Time is the obvious
answer in the shape of time to think,
consider and make decisions rather
than administer them.
That is an important part of the
philosophy behind the technology we
are developing at Ohpen. Collect &
Conclude is a good example because it
starts with something almost everyone
involved in mortgage lending will
recognise. Skilled and experienced
people still devote an extraordinary
amount of their working day to
gathering, checking and reconciling
information before they reach
the point where their skills and
experience are really required.
Collect & Conclude uses multiple
OCR engines, AI models and
configurable rules to extract that
information from documents and
reconcile it with other trusted
digital data. The technology can
therefore identify inconsistencies
and present a consolidated view of
the applicant rather than relying
upon an underwriter to discover those
inconsistencies manually.
The ambition is not to ask
technology to exercise judgement
for us. It is to give people beer
information and more opportunity to
exercise judgement themselves.
Machines can provide consistency,
but consistency should not be
confused with judgement. Technology
can help similar circumstances receive
similar treatment and make the
resulting paerns easier to examine
and challenge.
People remain much beer equipped
to understand context and recognise
when an exceptional circumstance
82
The Intermediary | October 2026
may justify an exceptional response.
The future is therefore unlikely
to be a choice between human and
machine judgement.
The opportunity lies in combining
the strengths of both. There is,
though, a complication that deserves
much more aention. How do people
acquire good judgement in the
first place?
Nobody arrives for their first
day as an underwriter carrying 20
years of experience with them. That
experience is accumulated gradually
and, historically, some of it has come
from doing exactly the administrative
work we now quite rightly want
technology to remove.
Underwriters have looked at
thousands of payslips, encountered
unusual transactions on bank
statements and discovered that
what an applicant initially said did
not quite correspond with what the
evidence showed. They have watched
simple cases become complicated
and apparently difficult cases reveal
perfectly sensible explanations.
Much of that work may have been
inefficient, but it was also part of
an apprenticeship.
Rethinking experience
That is certainly not a reason to
preserve unnecessary administration.
There is lile merit in asking people
to continue manually reconciling
documents simply because previous
generations had to do it. It does mean,
however, that measuring AI simply by
the number of hours it removes from a
process misses something important.
If technology removes the repetitive
tasks through which junior employees
once developed paern recognition
and experience, institutions will need
to find another way of developing
those capabilities. Perhaps the
underwriter of the future starts much
nearer the decision itself.
Rather than gaining experience
by processing 1,000 relatively
JERRY MULLE
is UK MD of Ohpen
straightforward cases, they might
learn by examining how experienced
colleagues deal with 100 exceptions.
Technology could show why a case
has been escalated, present the
information it has reconciled and
enable someone to interrogate that
evidence rather than merely accept
the answer produced for them.
In that sense, AI could become
part of the apprenticeship itself,
exposing people to a greater variety
of circumstances much more quickly
than traditional administrative work
ever could.
But this will not happen by accident.
We need to recognise the issue
otherwise we create a curious paradox.
We automate administration precisely
so that experienced people have more
time to exercise judgement, while
gradually removing the route through
which the next generation develops
the experience needed to exercise that
judgement well.
This is where the discussion about
AI in banking becomes much more
interesting than a conversation about
productivity. The questions become
organisational and, to some extent,
philosophical. What are machines
best placed to do? What should
remain with people? Where does
accountability belong? And what
experience must someone acquire
before an institution should trust
them to make decisions?
Opportunity is created when
technology takes on more of
the collection, validation and
reconciliation of information.
Underwriters can spend less time
pursuing paperwork and more
time understanding customers and
assessing risk. But giving people
more time to think solves only one
part of the problem. We also need
to make sure we continue teaching
them how. ●
