The State of AI in Business: Big Expectations, Slow Progress
This article will count 0.25 units (15 minutes) of unverifiable CPD. Remember to log these units under your membership profile.
AI continues to dominate strategic conversations in audit, accounting and advisory firms, but new evidence shows a widening gap between what leaders expect from AI and what is actually happening in day-to-day operations. A new survey from audit-tech provider Fieldguide highlights this paradox clearly:
94% of professionals say AI speeds up engagements,
90% say it reduces workloads, yet
88% still rely on manual data collection, and
80% still document by hand.
At the same time, the MIT State of AI in Business 2025 report paints an almost identical picture on a global scale: despite $30–40 billion in spending, 95% of organisations see no measurable financial return from AI projects, and only 5% manage to scale AI successfully.
Together, the two reports show that the profession is eager to adopt AI, but struggling to operationalise it.
Why Firms Are Struggling to Implement AI
The pilot-to-production gap
Fieldguide found that 81% of AI rollouts are abandoned due to lack of stakeholder buy-in. MIT calls this the “GenAI Divide”: companies run successful pilots, but can’t convert them into real workflow change.
Poor and inconsistent data
Both reports highlight data quality as the biggest practical barrier. Scattered evidence, inconsistent naming, and siloed systems mean AI can’t learn, remember or adapt, making advanced automation impossible.
Integration friction
Many AI tools are not designed for real-world firm workflows. MIT found that enterprise tools often “sit next to the workflow” instead of integrating into it, which leads staff back to manual processes.
The human challenge
Professionals are hesitant to change established ways of working. Fieldguide reports widespread lack of internal buy-in, and MIT confirms that user resistance, poor UX and unclear ownership prevent adoption.
3 Habits of Successful Firms
Both Fieldguide and MIT found that the organisations successfully using AI share three habits:
They choose tools that learn and adapt
These firms avoid generic “AI add-ons” and instead choose systems that can:
✅Remember past decisions
✅Adapt to workflows
✅Integrate deeply into existing systems
✅Improve with feedback
MIT calls these learning-capable systems, and they are central to the 5% of firms that see real results.
They build “experiment-friendly” cultures
Successful adopters:
✅ Start small
✅ Run quick experiments
✅ Create safe spaces for trial and error
✅ Document what works
✅ Share playbooks across teams.
This builds trust and smooths the change-management process.
They treat AI as a partnership — not a software purchase
Instead of buying off-the-shelf tools and hoping for the best, these firms:
✅ Co-develop solutions with vendors
✅ Customise AI for specific workflows
✅ Focus on business outcomes, not features.
MIT found that externally partnered implementations are twice as likely to succeed as internal builds.
What This Means for Accountants
For practitioners, the message is simple:
AI works only when data, people and workflows change with it.
Start small with practical use cases like document review, reconciliations and workflow automation.
The biggest wins are in back-office efficiency, not flashy front-end tools.
As Fieldguide’s CEO puts it: “The profession doesn’t lack belief in AI — it lacks follow-through.”
Source Article: Accounting Today
