The Intermediary – August 2026 - Flipbook - Page 77
P RO T E C T I O N
Opinion
Purpose before
technology
A
dopting new
technology is a
huge challenge.
70% of digital
transformation
initiatives fail to meet
their objectives.
If this is already true of familiar,
well-established systems, how much
more difficult will it be to master
artificial intelligence (AI), which has
vastly more potential, but also carries
new and unpredictable risks?
For anyone involved in advising
clients, AI will make its presence felt
in two ways:
First, it will appear within
intermediary firms, not only
increasing productivity by taking
notes and draing documents,
but also by performing some, or
all, of the analysis that creates a
recommendation.
Second, it will be used by clients –
from those who are feeding random
questions into large language models
to those who are using speciallydesigned ‘agentic’ AI to make and carry
out a range of decisions for them.
This year, the Chartered Insurance
Institute (CII) brought together
professionals from a range of sectors
to set out professional standards for
adopting AI in this sector. It began
with the principle of ‘purpose before
Often, it is
intermediaries that secure
justice [for example]
by challenging claims
decisions on life assurance
that have been made
without considering all
the facts”
technology’. Many organisations
inside and outside financial services
are adopting AI because they feel
they need to be seen to act, bolting
tools onto existing processes instead
of asking the harder question first:
where, if anywhere, does AI genuinely
serve the organisation’s strategic
objectives?
The guide looks at specific examples
where detailed, demanding standards
can drive good practice. For example,
on responsible deployment of AI it
sets out ‘red lines’. These state that
professionals should not use AI for:
Social scoring or behavioural
manipulation.
Emotion recognition for eligibility
or pricing.
Fully automated decisions that
significantly affect customers
without human review.
Any use that can reasonably be
expected to undermine customer
autonomy or trust.
Equally, when it comes to third-party
and vendor controls, the guide argues
that firms should ensure that:
All external AI tools undergo due
diligence, risk assessment, and
contractual controls.
Vendors provide transparency on:
data sources; the model logic that
sets out how AI learns and operates;
and performance – showing how
outcomes fit with the purpose they
have been designed to serve.
Controls and training are adequate
to avoid inappropriate use of
unauthorised tools
It also stated that when customers use
their rights to request an explanation
of how a firm’s AI system works the
system should not fall into technical
descriptions (“the model weighted
X and Y”) or generic statements
(“your risk profile changed”). Every
explanation should meet the ethical
requirement for clarity, relevance,
honesty and professional justification.
DR MATTHEW CONNELL
is director of policy and public
affairs at the Chartered
Insurance Institute
The guide sets standards for
intermediaries, but it should also give
advisers a tool that they can use to help
their clients.
Oen, it is intermediaries that
secure justice for clients – for
example: by helping a client avoid
mistakes prompted by ‘non-advised’
sources; by challenging claims
decisions on life assurance that have
been made without considering all
the facts; or by securing a mortgage
loan for a client with an unusual
risk profile.
Intermediaries can use these
standards to secure beer outcomes
for their clients – demanding that
providers use AI responsibly, and that
they communicate with customers
in a meaningful way that can be
questioned and challenged.
Better success
Intermediaries can also use these
standards to keep ‘agentic’ advice
services honest – measuring
their levels of transparency and
accountability and ensuring that they
have identified an acceptable purpose
– and are living up to it.
Understanding the professional
standards that all financial services
firms should meet will help
intermediaries in their mission to
secure the best deal for their clients
– and help to create an environment
where AI systems are adopted with a
beer success rate than three in 10. ●
August 2026 | The Intermediary
75
