On 7 November 2025, ISACA SG Chapter (ISACA SG) and Chartered A Accountants Australia and New Zealand (CA ANZ) Singapore organised our second roundtable discussion addressing the key concerns about artificial intelligence transformation.The roundtable was well represented by ISACA SG and CA ANZ invitees comprising senior leaders representing finance, technology audit, risk management, internal audit andcybersecurity to provide a holistic view ofthis discussion topic. Amongst the participants, there was a good mix of both financial institution (FI) and non-FI industry representation. The conclusion is that “AI projects can be seen as a strategic move if deployed right. The challenge lies in what’s “right” for your organisation and that requires huge effort to align properly”.
In today’s fast-evolving and fast-adoptiondigital landscape, organisations are expected to ride the AI wave to either boost its business performance or minimally boosting its productivity or both. AI is widely recognised for its potential to transform organisations. This is evident when private-sector investment in AI increased 18-fold from 2013 to 2022 and one survey discovered 58% of midsize organisations had deployed at least one AI model to production as cited by Ryseff, f, De Bruhl & Newberry, 20242.
Every transformation project experiences both positive and negative outcome and feedback. It is because any transformation project involves human p participationheavily and this brought on many sensitive challenges when dealing with the subject of human. Perhaps that is why when the statistics show that most transformation project failed is not a surprisingnews to some of us. On the other hand, some newer data suggests that organisations that treat AI as a strategic transformation (rather than just a technology experiment) report much higher success rates.
Do you think AI Transformation Projects have reached maturity or have met its purpose?
At the start of the discussion, the author (who is also the discussion moderator) asked the participants whether they think their respective organisations’ AI projects have been implemented successfully. More than half of the participants gave feedback that their projects are not quite successful (i.e. < 50% objectives met).
Given that majority of the participants concurred that their AI project implementations are not quite successful, the reasons and challenges of such low success rates were explored in the discussion. The following are a summary of the discussion points:
- Define success – the term “success” is subjective and hence, the need to have clear metrics to make it an objective subject. And precisely that most projects do not have metrics outlined prior project implementation, or for those that have metrics outlined in their approval papers but were not measured against the metrics accordingly.
- Start small or ready to fail – – some projects are prepared to fail at the beginning to enable better sensing of what success looks like. These may be included in the survey statistics and hence making the statistics a little skewed.
- Follow the AI adoption trend – many may have started their AI projects without the end in mind, they just follow the wave and trying to have some head start to hopefully gain some competitive advantage.
- Business returns of priority – there are some where purpose not met is fine as long as their customers accept the fact that AI projects have been deployed and perceived that the businesses must be of higher quality. To business owners (especially start-ups), hitting return on investments (ROI) is key and to demonstrate that there are room for improvements with assistance from AI projects to attract more sponsors/investors.
Do you think AI talents are readily available and what are the challenges in developing and retaining AI talents?
“World Economic Forum (Oct 2025)3 observed that today, 94% of leaders face shortages, with around one-third reporting gaps of 40-60% in AI-I-critical roles.”
Not every organisation can afford an AI talent like Meta did. So, for the many “normal” organisations out there, how and where do they hire their AI talents? A At this juncture, not many have seen Boards as AI projects sponsors as the projects’ business values have not been clear. Management as sponsors observed that the organisations embarking on AI training for their employees early. While not all employees are expected to step into AI developer role, they need to understand how to leverage on AI to do their jobs better. Hence, employees who cannot work with AI well may risk losing their roles. All participants agreed that competencies need to be developed over time. The reality is that the whole world is suffering from AI talent shortages. All the more this t temporary talent shortage provides many employees the opportunity and time to pick up such skills and transform their roles accordingly.
An ironic situation happened now that many professionals went ahead to obtain AI related certifications and called themselves AI trainers &/or AI talents. Many of them do not have handson experiences to claim the status. I It was unanimous from the participants that o one has to handson with AI projects to understand and gain the how-to to implement AI projects. Especially with GenAI projects, you will not know what to expect from the GenAI after you have developed its model with your data for learning. Many mistake AI projects as just another automation project. So, the way to do this, although primitive but it works, is to openly recruit people and expect to train them accordingly. Without the initial investments to train, the talent gap will forever be there.
Do you think culture play an important role in AI transformation projects? Will it make a difference in mitigating AI risks?
In the Workforce Ecosystems: A New Strategic Approach to the Future of Work research from MIT Sloan 4 noted that “strong internally focused cultures, resistance to change, and organizational silo behaviours can stymie workforce ecosystems”. Culture does impact risk taking (Rehbein, 2014).
This view of “resistance to change” was concurred by all participants. This is especially true when employees “gang up” to prevent their job loss probability and exhibited resistance to change made AI transformation projects difficult and costly. On the other hand, there have been the other extreme of “loving AI culture” demonstrated by some organisations as observed and experienced by some of the participants. They observed interesting behaviour from the organisations that placed too much importance and reliance on AI produced output such that these organisations seem to be in their own “hallucinations” state.
There is a saying that “culture eats strategy for breakfast”. Since culture can influence risk taking behaviour, organisations should always ensure that their work culture is healthy and always put human as priority over AI projects. Another observation made by the participants that in addition to culture itself, not many organisations can handle change management well.This is evident when Management decides on embarking transformation project did not factor in work re-designing as one of their change management strategies. Hence, most transformation projects generate fear in employees leading to the above-mentioned challenges and outcome.
Is security-by-design essential in AI implementation?
“Yes!” Not surprising that the GRC participants responded positively. However, our business participants said that such methodology and practice are transparent to them. By the time the projects are passed on to them for testing, this concept is too late to be considered. As the training and understanding of a project team about security-by-design differs, it is suggested to include security risk as a criteria when evaluating and developing any AI projects.
“A“According to the F5 Networks “2025 State of AI Application Strategy Report”, although 96% of organisations deploy AI models, only 2% qualify as “highly ready” to secure and scale these systems.”
Conclusion
Embarking on the AI transformation project journey is a continuous one. However, many still fail to learn from the mistakes that others have gone through. Regulations and standards are always lagging behind, hence, project sponsors and teams have to be mindful of not neglecting security and risk strategies as part of your project considerations. Regardless which technology evolution phases we are in, research has shown that every technology evolution project implementation encountered two-third failures. The statistics for AI project implementation is also telling us likewise. The word “transformation” will always include human-in-the-loop. However, how to manage the change process well depends heavily on the maturity of the organisations people and culture.
Author: D Dr. J Jenny Tan (ISACA SG)
Organisers
- CA ANZ – Samantha Chan, Acting International Leader and Daniel Ngo, CA ANZ Singapore Overseas Regional Councillor About CA ANZ (https://www.charteredaccountantsanz.com/)A professional accounting body | Chartered Accountants are known as Difference Makers
- ISACA SG – Dr. J Jenny Tan, I Immediate Past P President and Yap Lip Keong, President About ISACA (https://www.isaca.org/about-us)A community of IS/IT professionals in pursuit of digital trust | We are working to build a better digital world.
CA ANZ and ISACA SG would like to thank the following persons for contributing their views at the round table leading to this article formulation:
- Alvin Neo
- Christopher Lek
- Chin Shi Mei
- Lim Ee Lin
- Rajesh Laskary
- Richie Tan
- Sourabh Haldar
- Sujeewa Padmasiri
- Winnie Ang
References
1 GRC: Governance, Risk & Compliance
2 James Ryseff, Brandon De Bruhl, Sydne J. Newberry. The root causes of failure for artificial intelligence projects and how they can succeed. 2024. RAND Corporation
3 https://www.weforum.org/stories/2025/10/ai-s-new-dual-workforce-challenge-balancing-overcapacity-anddtalentshortages/#:~:text=Acute%20shortages%20in%20AI%2Dcritical%20talent,anticipate%2020%2D40%25%20gaps.
4 https://sloanreview.mit.edu/projects/workforce-ecosystems-a-new-strategic-approach-to-the-future-of-f-work/5 KPIs: Key Performance Indicators
