An Old Dog Learns a New Trick
My First Use of AI for a Work Assignment
Recently, the author received an assignment which called for him to make a calculation that he had not performed in the past. With great hesitance, he decided to ask AI for help. In making the leap to enter the AI pool, he found the answer he needed and confidence to “try it again.” He has found a new tool to put in his litigation support tool kit. In this article, he shares his experience.
I am 68 years old with a birthday coming up in October. This means next year I will turn 70 years old. I began litigation support work in 1997 after a career in banking and finance. Throughout my working career, I have had to learn how to use new technologies. When I started in banking, we were still dictating our letters to the typing pool. Then computers made it easy for us to type our own letters, once we learned how to use the computer. My first mobile phone was in a carrying case the size of many carryon bags for today’s air travel. When artificial intelligence (AI) began to be used in our field for research and a source for written reports, I became concerned. I am old enough to remember HAL in 2001 A Space Odyssey and of course, Arnold Schwarzenegger in Terminator.
My first presentation regarding AI was at one of NACVA’s national conferences. The session was very informative and scary. I vividly remember the speaker telling us that an autonomous car will more than likely be programmed to protect the most people in an accident situation. His example was set in Vegas (which is where the conference was). He stated that if your autonomous vehicle had such a program and three or four partiers walked out in front of your vehicle, it would choose to save their lives and crash you into a building. As you can imagine, I have never ridden in an autonomous vehicle.
More recent sessions I have attended focused on getting “good” usage out of AI. They have shown attendees how to use the various versions of AI to help research, organize financial data, and even with wording in reports. These sessions have been helpful and somewhat assuring for a non-user to think it would be “ok” to try.
The Assignment with No Answer
Two years ago, I received an assignment from a law firm representing a successful business in Fort Worth, Texas. The entity (ABC) had been in business for over 30 years and had grown to have locations across Texas and in other states. For over 10 years prior to my being contacted, ABC had worked with a computer service company (XYZ) for tech support. This computer service firm had provided or recommended the hardware (computers and mainframe) for the business, along with providing software, cyber security, and off-site and cloud storage.
ABC had a significant tech failure a year before I was called. The software programs did not work and efforts to bring their mainframe back online failed. Important in this failure was the loss of their e-mail accounts.
As with many businesses, once a person or business became a client, a specific person was assigned to manage the account. Most of these account managers maintained individual e-mail files for their clients, which contained copies of contracts, purchase orders, invoices, and other communications, which provided the history of the relationship.
When the system was brought back online, over 80% of ABC’s e-mail files were gone. A thorough review showed that over 90% of their business files in other areas were lost.
ABC asked XYZ to provide copies of its backups to assist in rebuilding their historical records. Unfortunately, XYZ had not maintained updated hardware or software. Its system would not link with newer cloud storage systems, and they had, without informing the client, stopped off-site storage in prior years. In other words, there were no backed up records from which to rebuild ABC’s files.
An outside tech firm was hired, which ultimately replaced the antiquated hardware and, through its efforts, was able to restore approximately 15% of the lost files. Unfortunately, the majority of the restored files were related to the operation and management of the prior software system being used by ABC.
This set of facts brought about a lawsuit in which ABC claimed economic damages caused by XYZ. In the lawsuit, ABC claimed a loss due to additional hours worked by employees trying to recreate the lost files and maintain the relationships with their customers. A second claim was for the value of the lost files.
Additional Hours Worked
The first calculation I was asked to perform was the loss from extra hours worked by ABC’s staff. This was a straightforward calculation. The entire staff of ABC, including the owners, put in extra hours working on recreating files. While these efforts never fully restored the files, it was an important effort in retaining their client base.
ABC’s office manager sent an Excel spreadsheet to each employee asking them to note the time spent when working on recreating a file or having problems with the software system (while it was being rebuilt) and had to do extra work to complete a contract or order.
Most of the employees were very detailed in their reports citing hours worked, which accounts for which the work was done, and the type of work. From this information, I was able to develop a spreadsheet with four months’ worth of daily data of additional work. Based on the pay rates for the various staff members, I was able to use the information to estimate the economic loss (out-of-pocket expenses) for that four-month period. In addition, I was asked to project the loss for three additional months for which data had not been gathered.
For the additional three months, I interviewed part of the management team regarding the extra efforts made the two months after the initial “crash” and in the final month prior to the new system being brought online. Using the data from the employee reports and the management team’s interviews, I projected the losses for those months.
A report representing this calculation was submitted prior to valuing the lost files. The attorneys representing XYZ complained that I did not have data for the three other months. They complained I took the employees at their word without outside verification and that there was no guarantee the employees would not have used that time for working on other projects so that ABC’s costs would have been the same. In the end though, the push back on the calculation for this damage claim was relatively light.
Value of Lost Files
XYZ’s attorneys and insurance adjuster pressed ABC and its attorneys for a value of the lost files. It became obvious the largest economic damage would be the value of these files. By this time, the new system was “up and running.” A review of company data confirmed that approximately 67% of e-mail and data files were lost.
Having never made such a calculation, I sought assistance from financial and forensic literature. Most litigation support literature was silent. Financial literature also offered no solutions. I asked the attorneys representing ABC to do a case law search for decisions valuing computer files. They found no cases that had set a precedent for such a calculation.
Finally, I turned to Google Scholar. Using various search topics (e.g., valuing computer files, the value of lost computer files, the value of lost data, etc.), I found more than 100 articles on valuing lost data and/or files. I reviewed 100 peer reviewed articles most closely relating to my assignment. The ultimate result was no one has yet to “figure out” how to value lost data files or computer files. I cited four of these articles in my report.[1]
It was obvious that, to determine a value for the lost data files, I would need to create a viable formula. Before doing that, I wanted to make one last effort to make sure I have not missed something, and this is where AI came in.
At a professional conference early in 2024, I attended a session on AI. The speaker addressed using a number of various AI systems but relied on ChatGPT for his examples. He emphasized making your question to AI as specific as possible.
I believed my AI search would confirm what I already knew; that no formula exists. Of course, my search might show me a source that I could use. So, I took a chance and became a member of ChatGPT. I joined at the lowest level possible; the one that provides the least privacy and does not charge a fee.
The question I put to AI was this, “I need to value computer files that have been lost due to an IT error. Are there any formulas that can be used to value computer files lost to data problems?”
Within five seconds, I received a two-page answer. The opening paragraph stated, “Valuing computer files lost due to IT errors can be challenging as it involves considering several factors, both quantitative and qualitative. Ther is no single formula, but a combination of approaches can be used to estimate the value of the lost data.”[2]
The answer provided a brief discussion for using the cost-based approach, the income-based approach, and the market-based approach. It also provided an example in which the cost-based approach and the income approach results are totaled. This total is then combined with additional costs (e.g., lost revenue per day).
In its conclusion, ChatGPT stated, “While there is no single formula to value lost computer files, combining cost-based, income-based, and market-based approaches, and considering direct, indirect, intrinsic, and legal factors can provide a comprehensive estimate of the value.”[3]
In my report section “Value of Lost Files,” I quoted this ChatGPT answer and one of the Google Scholar articles. These two sources confirmed that there was no single formula or method for valuing lost computer data files. The next question was, “How do I value lost files for this assignment?”
The Solution
Initially, I discussed valuing the lost files using a willingness-to-pay model. I asked the owners of ABC, “What would they be willing to pay or have insurance pay if hackers took control of your computer system and demanded a ransom to release the files they controlled?” That idea was quickly shot down. The owners stated that event had already happened and they did not want that information made public nor would they discuss the ramson amount paid to release their files.
While discussing other options, I found that ABC paid the account managers 15.5% of their clients’ gross revenue as a bonus for retaining that client another 12 months. The owners agreed that between 3% to 5% of the gross revenue bonus was for building and updating the historical files for that client.
Because this had been ABC’s long-term method for paying the account managers for the retention of a client and updating the historical files, I felt it provided a way to value the files.
I asked for and received the gross annual revenue for the lost files for the four full years prior to the system failure. This revenue included clients whose relationships existed long before this period, began during this period, or ended during this period. Therefore, it included turnover and retention.
From the annual gross revenue, the 15.5% bonus amount was determined. I then estimated the value of 3% of the gross revenue, 4% of the gross revenue, and 5% of the gross revenue. These values were shown as a range of value for the files and the loss for ABC.
This case is ongoing, which is why no actual names have been used in this article. The defense attorneys have not sought my deposition. I know both sides are in negotiations to settle the case.
Conclusion
Making the decision to ask AI to assist in my research for an answer to one of my problems seems like an easy thing to do, but without having done it before and not having great confidence in what AI will provide, I had to be willing to take a chance. The results gave me more confidence to use AI, which I have done with other assignments.
I am not an AI guru and appear more like a neophyte among neophytes. However, I have found a new tool to put in my litigation support tool kit. The only recommendation I can give is, when you decide to use ChatGPT or any of the other AI sources, be as specific as possible in your request. This will limit peripheral data from being provided. And if you find your results too limiting, expand your question.
I found you can teach an old dog at least one new trick. But to do that, I had to convince myself it was worth the leap.
[1] The four articles were: “The Market Value and Reputational Effects from Lost Confidential Information”, Joseph Tanimura, Eric Wehrly, draft May 2009, SSRN, www.ssrn.com//abstract=1083891; “A Review of Data Valuation Approaches and Building and Scoring a Data Valuation Model”, MITRE Corp, Harvard Data Science Review, 2023, Public Release Case Number 21-3464; “Data Valuation and Law”, Jordan Barry, D. Daniel Sokol, Southern California Law Review, Vol. 96, 2023; “Valuing Data as an Asset”, Laura Veldkamp, Review of Finance, Oxford University Press, 2023.
[2] ChatGPT response 06/05/2024.
[3] Ibid.
Allyn Needham, PhD, CEA, is a partner at Shipp Needham Economic Analysis, LLC, a Fort Worth-based litigation support consulting expert services and economic research firm. Prior to joining Shipp Needham Economic Analysis, he was in the banking, finance, and insurance industries for over 20 years. As an expert, he has testified on various matters relating to commercial damages, personal damages, business bankruptcy, and business valuation. Dr. Needham has published articles in the areas of financial and forensic economics, and provided continuing education presentations at professional economic, vocational rehabilitation, and bar association meetings. In 2021, Dr. Needham received a NACVA Outstanding Member Award. He is also a member of NACVA’s QuickRead Editorial Board.
Dr. Needham can be contacted at (817) 348-0213 or by e-mail to aneedham@shippneedham.com.
