Why leadership matters when implementing AI

Organisations are investing heavily in AI, yet few are turning experimentation into measurable business value. Steve Durbin, Chief Executive of the Information Security Forum, argues that success depends on leadership staying closely involved in priorities, guardrails and results.

A fleet is only as fast as its captain is wise. AI adoption works the same way, when guided by leadership that understands the mission. AI is already reshaping how work gets done. Yet despite widespread piloting and experimentation, many organisations are discovering that AI alone doesn’t absolve gaps in leadership or create immediate value.

Here’s a sobering reality: most companies are testing AI in some form, yet only about five per cent have turned it into revenue gains and cost reductions, and 60 per cent say it has made little to no difference so far.

The difference between AI that delivers value, and AI that stalls in pilot mode, really comes down to leadership.
“Leaders must look for meaningful AI use cases”

Extension of human capability should be the key driver of AI adoption. The idea is to enable outcomes that weren’t previously feasible. Unfortunately, the inconvenient truth that leaders barely acknowledge today is that AI is still such a mystery, still not easy to comprehend.

The disconnect is clear. According to a Gartner CEO survey, fewer than half of CIOs are truly AI savvy. Although aware of it being the next frontier of technology, they know better than to be seen obstructing AI deployments. A gap exists between what the business truly needs and how AI can fulfill that need.

Leaders must think of how AI can realistically help their business succeed. An example use case could entail the development of an AI agent that can digest all call notes, emails, and CRM updates into a unified customer narrative, allowing a salesperson to pick up from where another left off, without looking for context. While agents are built upon programmed Large Language Models (LLMs), creating them means a focus on defining goals, connecting tools (and stakeholders), providing instructions, and setting constraints: much in the same way that leadership should govern.

“The hype is real, but look beyond it”

There is always a chance that leaders are being bowled over by a great demo that answers the question: “Can it do this?” But leadership must give answer to the harder question: “Will we run the business differently because of it?”

Further questions can help clarify purpose and intent behind AI adoption:
• What decision, workflow, or customer moment will change next quarter?

• What are we willing to sacrifice or stop doing, even if it’s familiar?

• What do we want to accelerate, and where do we want friction on purpose? (e.g., compliance, brand voice, approvals)

• What would ‘good’ look like in numbers, not just adjectives?

Too many companies adopt AI defensively, driven by board pressure, competitive anxiety, or the pretence of needing “an AI story” to promote.
T his produces pilot programmes that don’t move the needle or reach satisfactory outcomes.

Leadership should shift the conversation from “we
need AI” to “we’re changing this part of the business,
and AI is how we’ll support it.”
“Executives should own AI
implementation”
AI initiatives rarely fail because the tech isn’t good
enough. They fail because leadership steps back
too soon. When the C-suite delegates AI entirely
to technical teams, whether from board pressure
or FOMO, projects may drift into “impressive
pilots” with no adoption or measurable impact. AI
implementation is not similar to your standard IT
rollout; it can change workflows, expose risk, and
demand accountability. Executives should therefore
be involved in setting priorities, defining guardrails,
and determining key metrics that define a successful
AI deployment.
Tech teams speak about capabilities. Business
teams speak about consequences. The CEO’s role is
to translate trade-offs and create a shared contract
that clarifies where AI can assist, where humans
must decide, and what evidence is required before
autonomy is granted. In short, if it can’t be explained
and measured in business terms, it isn’t ready for
prime time.
“Manage employee anxiety and
adoption risk”
Leaders can get everything right, but if they don’t
address employee concerns around AI adoption,
they can get trapped in a vicious cycle of AI pilots.
Employees must buy into their changing roles as
AI gets embedded across workflows. The fear is
real, and a good leader will aggressively address
employee concerns upfront. This can only happen if
workers are well aware of what AI cannot do. There
should be an understanding that human oversight
is required. AI cannot be viewed as ‘implement and
forget’. Its operational impact needs to be reassessed
regularly, and as trust increases, so should guardrails
be reimposed.
Not enough leaders are debating the merits
and challenges of AI. The focus must be on
adopting AI with enough discipline to create
measurable value. AI creates advantage only when
leaders own the mission and stay accountable for
the trade-offs. ■
Further information
To find out more about the ISF’s cybersecurity
insight, risk guidance and leadership resources, visit
www.securityforum.org