Is it right to use AI in preparing a valuation?
March 5, 2024
Some interesting findings from BVR in the US shines a spotlight on the issue
We all know that AI is looming. Indeed some of you have already built AI (algorithm-based) solutions that are assisting you in areas like due diligence.
(On the issue of due diligence, progress our end is moving at pace so watch this space for more news).
As valuers we keep a close eye on how AI will be introduced to valuations work. We remain ultra cautious on the issue until we start hearing that HMRC and the courts will accept valuations based on AI. Use of it until then, excepting only in a vanishingly rare number of situations, is to us just too risky.
One of our mantras is to produce reports where the chance of challenge by the authorities is de minimis.
Back to the BVR research…
BVR published this article last week: Unexpected results from survey on AI usage at BV firms | Business Valuation Resources (bvresources.com)
Rod Burkert of Burkert Valuation Advisers in Colorado Springs, a leader in business valuation data and research, undertook some research of BV firms on their use of AI.
Of those he polled, Burkert says 48% “are using artificial intelligence in their practices, despite a lack of ethics guidance from the valuation professional organizations (which are working on it), Burkert notes. The usage is mostly for researching economic or industry conditions (see below). “This (I think) is a very low-risk application and a great way to get familiar with AI,” he says.”
| How are you using AI in your practice? | |
| Research of economic or industry conditions | 61% |
| Using as an “editor” for valuation report narrative | 43% |
| Assist in marketing efforts | 39% |
| Assisting with analysis of comps for the market approach | 26% |
| Analyzing historical financials of a subject company | 17% |
| Examining financial records for forensics purposes | 4% |
| Other (please specify) | 30% |
The “other” category includes legal research (tax code, court cases), coding, and compiling public stock data (but without much success yet). One firm is thinking about building and deploying an internal ChatGPT application to analyze past reports and Excel models to help draft valuation reports.
We have used Chat GPT to try and uncover information we cannot readily find elsewhere, but only as a research tool. If Chat GPT tells us anything we then separately find the verifiable source before using the information.
I’m not sure I would yet trust AI to produce valuation report narrative. The cases are just too unique and when we have run tests the use of English is not good enough to use without editing. Our experienced valuers are still better and faster at writing the narrative right first time.
When it comes to comparables, the sources we use already have sufficient analytics in place for us to interpret the data speedily. The team also enjoys this part of a valuation exercise as it exposes them to much wider areas of business analysis. They learn while doing, so it would be destructive to take this away from them.
Our tech team has found Chat GPT useful for giving us scripts for software code from time to time.
AI vs humans
There is so much chatter about how AI will take away white collar jobs in the way that the Industrial Revolution (in its many phases) has removed blue collar jobs.
Right now I am not a believer in this thesis. For quite a long time, it will be possible to introduce AI technology to improve mundane tasks. Algorithmic solutions will enable high volume low value work currently done by humans to be done better, faster and cheaper. But it will take longer than one thinks for the highly valued work that is subject to peer review, to be replaced by a machine.
… And on the way we have to think about the humans of the future. Training our teams by showing them the how and the why will be crucial to them joining us at the highest level as soon as possible so they contribute positively to our businesses.
I love the analogy of a radio. Yes we all learn by using the information it imparts, but everyone else can access that information too, so there is not that much of a premium on it. AI is a bit like this at the moment in terms of use in areas relevant to us. (I’m not an expert on its application in other industries).
The people who make the focused money out of a technology are the ones who know how to design, build, sell and crucially own the IP for the next generation of product, supported by their advisers who make sure they get the legals right!
We will all benefit from supporting the creators of AI solutions. We have already done a number of valuations for companies with this type of IP.
Yes, more solutions are arriving into our markets that will help us to be more efficient and in turn make our businesses more efficient and profitable.
Our valuation technology means we can do a complex valuation in a few days rather than weeks, but the inputs are still driven by and the outputs analysed by humans with experience and understanding. An AI valuation tool would have to be trained on thousands of reports before it could come anywhere close to doing an acceptable job in our view and frankly where are the software developers going to find the reports on which to train the AI?
We remain mindful that our wonderful team must remain interested and curious so they ask the right questions and get you and your clients the right answers. So would I let a robot replace them? Definitely not. Each and every one of them is cleverer than today’s generation of AI machines.
And as you know, we work fast so the client can get a valuation quickly, whilst comfortable in the knowledge that experts have done the work, with PI cover in place.
