Generative AI: More bane or boon for private market investors?

It’s incredible to see how quickly generative AI has proliferated as a mainstream topic over recent months, first with last November’s launch of the now-ubiquitous GPT3.5 chatbot incarnation ChatGPT, then with Google’s less-than-impressive demonstration of conversational AI service Bard, and lately with Microsoft’s announcement of its GPT4-powered in-program assistant Copilot for its Office suite of applications.



It’s incredible to see how quickly generative AI has proliferated as a mainstream topic over recent months, first with last November’s launch of the now-ubiquitous GPT3.5 chatbot incarnation ChatGPT, then with Google’s less-than-impressive demonstration of conversational AI service Bard, and lately with Microsoft’s announcement of its GPT4-powered in-program assistant Copilot for its Office suite of applications.

For the uninitiated, generative AI refers to a type of artificial intelligence based on deep learning and natural language processing models that can create new and original content, such as images, videos, or even entire texts, without being explicitly programmed to do so. This technology works by analysing and learning from existing data sets, and then using that information to generate new content that is similar or related to the original data.

As a private equity investor, I’ve been curious about the potential in harnessing generative AI’s capabilities to execute and, dare I say, improve the investment process. The myriad of workstreams involved has traditionally been carried out by throwing man-hours at each task – studying historical performance and analysing market conditions, then turning the conclusions into cash flow projections and valuation benchmarks, eventually producing returns analyses that may or may not confirm initial hypotheses, by which point significant resources would have already been invested.

Here’s why I think generative AI will be a game changer when fully (and properly) leveraged.

Enablement of sheer analytical power

Investing into mature companies often means that one is evaluating a business that is unlikely to be a simple singular product/service offering, but a multi-faceted, multi-market platform serving a diverse customer base along some level of bespoke pricing models. This effectively means that the historical datasets that need to be unpacked and scrutinised would be reasonably complex and potentially structured differently to what is familiar to the investor.

While today’s methods involve teams of analysts spending weeks meticulously poring and parsing through said datasets, generative AI combines tireless computing power with natural language models to produce observations interpreted from the data that is easily interpreted by humans.

With GPT technology being worked into common analytics tools like Google Sheets, it is now simpler for investors to access something so powerful to identify potential risks and opportunities. Chris Farmer, founder of SignalFire, explained how his firm has traditionally used AI to collect data on targets and benchmark their performance but is now using generative AI for more granular analyses:

"The new large language models are now able to understand a lot of those things," Farmer said, referring to AI abilities to interpret vast amounts of company information. "So rather than doing a bunch of coarse filters, we can actually just throw [the data] into an AI, and we're using these language models to understand the connectivity between and the similarities [of companies] quite accurately, with fewer layers of process."

Would we be able to simply ask an AI platform “hey, which company should I invest a million dollars in today?” in the next few years and get a reasonably reliable response? Probably not, but for a data-driven investor like Aura Private Equity, the unlocking of machine-enabled compute in what was previously a human-driven process is a particularly valuable advantage in ensuring insights are derived from collectively exhaustive analyses in both an efficient and timely manner AND without noticeable influence of human error.

Flexibility in analysing across different data characteristics

On several occasions, I’ve received financial and operational information from counterparties that are not particularly easy to digest. I’m not averse to rearranging things to make them more legible, but when the words are in another tongue altogether, paying attention to language class back in school starts to sound like a great idea. Okay, let’s use online translation, problem solved. Then what do you do when your spreadsheet has financial numbers in different currencies?

When one’s investment focus is knee-deep in such a diverse region as Southeast Asia, one comes across issues like these on the regular – varying languages, numerical syntaxes, currencies, and so on. Short of becoming aspirational polyglots or mathematical savants, investors can use generative AI to read data and produce outputs regardless of what form the inputs came in. With optical character recognition, the documents fed in can be scans or even pictures taken with a smart phone, instead of spreadsheets and machine-encoded text, which can be a great help when working with markets that still rely on older documentation methods.

Financial and operational forecasting led by internal and external data

Much of a private equity investor’s work is to form an idea of how far a business is expected to grow within defined and undefined timelines. It’s not that we have embraced our omnipotence in divination, rather we do it for very practical reasons, for example to set incentive targets for key management personnel or to structure deferred payments to a seller for whose company we’re buying. It is naturally important then that we have a very clear idea of how we can, well, “predict” the future.

The problem is that several investors often do not have a very clear idea, or in fact a clear idea at all. If a company has grown revenues by 50% a year for the last two years, we can perhaps say it will do that again this year for a hat trick (or can we?). Heck, let’s add a 20% haircut and call it a day! Most of the time, investors work in certain principles to their methodology in the interest of loss minimisation, such as discounting assumptions or referencing past sales and profitability performance. Generative AI takes that one step further by generating forecasts using not just one, two or even three historical metrics, but as many as your dataset allows.

Another aspect which is incredibly useful is the AI’s ability to use data from not just the data you have on the company, but the entire Internet, to aid in the process. Imagine if it was possible to project a Vietnamese F&B business’ sales in the next twelve months by amalgamating studies of past currency movements, consumer behaviour survey results, macroeconomic markers like disposable income and cost of living trends, and so on, in addition to corporate data – all using a technology that silently and persistently carries out the computing without weariness or error in a much shorter time.

Of course, alongside extolling the virtues of generative AI, we must also be mindful of the deficiencies (yes, there are weaknesses). For instance, there is an unavoidable potential for bias – generative AI algorithms learn from existing data sets, which can be biased in various ways. If the data sets used to train the AI algorithm are biased, the algorithm itself may also be biased, leading to inaccurate or incomplete analysis.

Another issue is the lack of transparency in generative AI algorithms. Unlike traditional investment analysis methods, which can be audited and reviewed by human experts, generative AI algorithms can be difficult to interpret and understand. This can make it challenging for private equity investors to evaluate the accuracy and reliability of the algorithm's output.

Generative AI algorithms can be vulnerable to manipulation or exploitation. Hackers or other malicious actors could potentially tamper with the data sets used to train the AI algorithm or manipulate the algorithm itself to generate false or misleading information.

Still, I’m looking forward to the new age of investing made possible by artificial intelligence. After all, it’s the closest thing to human-based investing.




This information is for accredited, qualified, institutional, wholesale or sophisticated investors only and is provided by Aura Group and related entities and is only for information and general news purposes. It does not constitute an offer or invitation of any sort in any jurisdiction. Moreover, the information in this document will not affect Aura Group’s investment strategy for any funds in any way. The information and opinions in this document have been derived from or reached from sources believed in good faith to be reliable but have not been independently verified. Aura Group makes no guarantee, representation or warranty, express or implied, and accepts no responsibility or liability for the accuracy or completeness of this information. No reliance should be placed on any assumptions, forecasts, projections, estimates or prospects contained within this document. You should not construe any such information or any material, as legal, tax, investment, financial, or other advice. This information is intended for distribution only in those jurisdictions and to those persons where and to whom it may be lawfully distributed. All information is of a general nature and does not address the personal circumstances of any particular individual or entity. The views and opinions expressed in this material are those of the author as of the date indicated and any such views are subject to change at any time based upon market or other conditions. The information may contain certain statements deemed to be forward-looking statements, including statements that address results or developments that Aura expects or anticipates may occur in the future. Any such statements are not guarantees of any future performance and actual results or developments may differ materially from those projected in the forward-looking statements. This information is for the use of only those persons to whom it is given. If you are not the intended recipient, you must not disclose, redistribute or use the information in any way.

Aura Group subsidiaries issuing this information include Aura Group (Singapore) Pte Ltd (Registration No. 201537140R) which is regulated by the Monetary Authority of Singapore as a holder of a Capital Markets Services Licence, and Aura Capital Pty Ltd (ACN 143 700 887) Australian Financial Services Licence 366230 holder in Australia.

Similar posts

Get the Latest News & Insights from Aura Group

Subscribe to News & Insights to stay up to date with all things Aura Group.