Tackling the new challenges of AI at the Contract Management Forum
This year’s Contract Management Forum took place on 23 September 2025 during FIDIC’s Global Infrastructure Conference which was held in South Africa. Moderated by Cosmin Tobolcea, a FIDIC Board member, and Thanos Totsikas, the FIDIC Credentialing’s (FCL) General Manager, the event was attended by practitioners who flew in from different countries spanning the four continents. The discussions were lively, and it was evident that there were many challenges arising from the increasing reliance on new technologies in the construction and engineering sectors.
Revolving around the theme of artificial intelligence (AI), the discussion centred on the challenges and consequences of the growing reliance on this new form of technology in contract administration and its effects on risk mitigation.
The speakers included Denis Serkin, Yuting Chen, Edulo Batalha, Andrea Chao and Cagdas E. Bayrak, who each shared specific examples of challenges faced, as well as some of the benefits of embracing the use of artificial intelligence (AI). They also spoke about how AI is transforming processes such as performance tracking, claim management, procurement, and dispute resolution—while also highlighting the risks and limitations that must not be ignored.
As artificial intelligence (AI) continues to reshape industries, its application in contract administration is gaining momentum. Arguably, the problem is not always the AI itself, rather, how it is used or collated. Serkin for instance emphasized the critical importance of data integrity, noting that “bad data means bad results.” At the same time, AI has the potential to accelerate and also facilitate the way we work. Design, for example, takes a lot of time to develop and finalise and often comes with many challenges.
It is common practice these days to rely on this technology as many believe that AI systems rely on high-quality, structured data to generate meaningful insights. However, in contract administration, this means ensuring accurate performance are part of the end result, while maximising tracking and payment records. For further accuracy, any results should be compared to real-time models to reduce subjectivity and potential conflicts. Yet, given the costs and safety risks associated with data collection, the software used should only be the final step to complete this. What must come first should be non-ambiguous questions and reliable inputs.
It is also important to understand how to manually process claims management before integrating AI, otherwise, the person using it may face issues as if they become insensitive to how the risks are allocated. This is because AI tools may misinterpret or overlook critical details without proper review or supervision. This underscores the need for human oversight and thorough contract review, even when using advanced tools.
Relying solely on AI-generated content without verifying sources is therefore dangerous. In today world, where speed and time are invaluable assets, the use of AI in legal research, particularly in preparing case law has increased significantly – but not without risks.
While AI can streamline legal workflows, it can pose potential dangers. Some lawyers have faced disciplinary action due to overreliance on AI as there was little to no verification conducted on the information it presented. Lawyers have been known to use AI to prepare caselaw and then been disbarred for not checking what they used AI to generate against caselaw and or legal precedent. Much was said about the need to read and re-check the contract after AI is used. Effective AI implementation therefore requires identifying corresponding elements in contract documents and understanding how they flow or confirm data.
While there may still be several setbacks or challenges, many believe that AI is here to stay. In particular, in the procurement processes, it can help prevent human errors at the proposal validation stage. AI also has the potential of bridging the gaps between information and capacity, particularly in regions where there is limited digital infrastructure. However, successful implementation requires tailored platforms that address local needs, especially in Africa and other developing regions.
AI is a useful tool when it comes to claim management for clients, but it must be clear how to do so manually before AI is used. The individual must look for corresponding elements in the documents then weigh this against how risk is allocated in the project.
The safety or data integrity aspects such as the cost of collecting data and errors (were considered and weighed against the General Purpose Technology (GPT).
Performance tracking and data comparison, i.e. the model versus the data received on a real time basis would be essential as this process should help remove conflict and make the data as objective as possible.
It was generally agreed that AI use is here to stay and that a future without it is not possible. AI offers a transformative potential for contract management and the potential to also change the subsets of dispute management. However, its success depends on data quality, human oversight, ethical use and tailored implementation strategies. For this, it will be necessary to adapt and upskill its application at the early to the advanced stages of the project. While there may be human errors in some of the validation processes for the proposals, the use of artificial intelligence or similar technologies should not completely rule out the human element. The best way to use AI is the for person who understands how it is used to use it.
The event was closed by James Mwangi, a FIDIC Board member who thanked all for a vibrant discussion. Attendees also expressed a strong interest in attending more events of this nature in the near future and encouraged FIDIC Credentialing not to wait a year to host another Contract Management Forum.
