Peyton Aldrich
To keep up with the risks and possible disadvantages of the use of artificial intelligence in construction, Enr resorted to two construction lawyers with a lot of experience in the industry and with issues related to AI: Christopher Horton, a member of Smith Currie Oles Llp and Peyton Aldrich, associate of the firm. They talked to the Elaine Silver correspondent about possible inaccuracies, such as sharpening consultations and what they call a “defensible path”.
What is the most important thing that construction companies need to know about the use of the AI?
Horton: Treat AI as a scalpel and use it strategically for specific inefficiencies, not as a capture solution. Take the time to carry out due diligence before adopting and to understand the safety, capabilities and limitations of the tool.
Aldrich: Do not forget that AI is a tool, not a replacement of human supervision. Without appropriate controls, it can be based on inaccurate outputs that could cause expensive errors.
There are many strengths and benefits, right?
Horton: Oh shines in streamlining projects planning, such as making offering stories or reviewing proposals to detect improvements. Our clients also use it to take notes at meetings, although safety must be guaranteed. It is important to veto safety tools and to implement clear guidelines for maximizing efficiency.
So that means what?
Aldrich: Companies must train staff to use the AI critically and establish policies to prevent ad hoc. Define tasks to automate, carefully evaluate tools and provide training on data entry and detection errors. Here is a personal example: I use the function of Westlaw Ai. Therefore, when I am starting a research project, if it is not a problem, I have a lot of knowledge of what I will do is that I will really take the directions that Westlaw gives us. This shows us, well, that’s how our IA works. This is how to ask the best kind of question and I will copy and paste to Chatgpt, and then say -this is the problem I am trying to look for.
Therefore, I use an IA to help -to generate a question to ask me the most accurate answer of another.
What are some of the largest obstacles?
Horton: Understand what the security features are and whether it is the closed loop or the open loop. It is very important not to put any information that includes any confidential property lawyer, client or work product or privileged information.
Christopher Horton
Aldrich: And, ai can generate skewed or inaccurate outings, erroneously interpret the contract clauses and hallucinations, where AI produces false information, such as cases or regulations of makeup. This could lead to, for example, choosing the winner of the wrong offer due to defective data, creating new liabilities. It is a real risk if you do not verify with the results carefully.
Why is the construction industry especially vulnerable?
Aldrich: The construction implies complex contracts and decisions of great participation, so that the errors of the AI, such as the wrong interpretation of a contract disposal, can lead to lost deadlines or financial losses. The construction is vulnerable because it deals with long contracts, personalized and heavy data processes, such as risk assessments and subcontracting evaluations. The tendency of the AI to overlook the terms of the nuanced contract or produce skewed outputs of incomplete data may have serious consequences, such as problems of fulfillment or unfair decisions.
Horton: The rapid nature of the industry tempts companies to trust AI without verification, especially for offers or schedules. But non -verified outings may violate contractual obligations or expose ownership data, especially if AI tools are not secure.
So, to train the IA more effectively …?
Horton: Training involves learning the limits and security functions of the tool. It is not just about using the AI, but to know when to trust it and when to verify it, which comes from structured orientation and implementation based on policies.
And what is a defensible path that you mentioned to me?
Aldrich: A defensible path documents how the IA was used, including the entrances, outputs and human supervision. This record reinforces your position in litigation or regulation reviews, demonstrating consistent and appropriate procedures.
That is Important because …?
Horton: Without a defensive trace, you cannot justify AI -based decisions in court or arbitration, especially if privileged information is committed. Opaial algorithms also make it difficult to defend decisions in disputes that may lead to an increase in legal and financial exposure.
What IA policies do you have in your company?
Horton: Our company has a strict Policy of IA that I developed as chairman of the AI Committee. We only use approved tools such as Westlaw, Chatgpt and Microsoft Copilot, ensuring that they are a safe and closed loop to protect confidential data. There is no privileged or privileged information in non -verified tools, and we are trying the AI generative as the Thomson Reuters cocounsel for future use.
What is the only thing all the construction executives do to use AI programs well?
Horton: Verify, verify, verify.
