AI revolution: What will it mean for private market investors?

The impact of the AI revolution on private markets investing

There are a few industries where the immediate impact of Artificial Intelligence (AI) will be massive. For some businesses, AI is already completely change how they operate.

For most other industries - including private market investing we believe that there will be a productivity boost felt by the nearly all professionals within the next six months. But that AI will not completely revolutionise what is being done. 

For now.

Large Language Models and their foundational property

Large Language Models (LLMs) are the subset of AI that is being most widely used in private market investing today. These models contain billions of parameters and have been built using copious amounts of text data from the internet. They are excellent at human language problems such as generating text, summarising and translation.

A key aspect of these models is their foundational property; this allows the broad purpose language models that have been built by the large tech players to be fine-tuned to solve specialised problems. Importantly, this fine tuning requires only a fraction of the compute and data power of the original models.

How will private investors use AI?

On-desk tools – such as Schroders in-house chatbot “Genie” – will drive the types of AI productivity gains felt by most people. This initial stage of the revolution should be thought of in the way computers or spreadsheets changed industry.

Writing support, presentation design, summarisation and coding are being integrated into essential business tools and are already proving their value.

When foundation models are paired with other functionality and propriety data, this can have a multiplicative affect on their value; creating novel solutions to time consuming tasks. The pairing of such models with other functionality (such as the ability to do calculations) and with proprietary data (such as internal documents) significantly reduces or even eliminates the risk of so called “hallucination”, which is the tendency of large language models to sometimes make-up answers.

Private investors are faced with enormous amount of information; general partner(GP) reports, company filings, industry papers, news articles, and market intelligence. The mass-extraction of this unstructured information, facilitated by LLM, is allowing investors to cut through the noise and focus on the most important pieces of information.

A long standing challenge for PE investors is building a “comps” list for the purposes of valuation. Traditionally, this has relied on sector classification and geography of operation to identify peers for an investment; now using the contents of companies’ websites in conjunction with LLMs its possible to build a more nuanced “similarity mapping” that creates a more representative list. And in record time.

Soon we will expect to see AI powered data rooms, where an AI assistant has access to all of the documents and information inside the data room and is able to quickly summarise the contents, answer questions and even highlight the most pertinent points. This will make due diligence efforts even more comprehensive, increase speed and save time

How will investors, fund managers, investment firms, and service providers adapt?

The organisations poised to lead tomorrow's market are those that can master the engineering challenges of integrating foundation models with internal data, that can quickly educate and empower their workforce to use AI as a productivity and that, foster a culture of innovation across the entire organisation. Additionally, the organisations that form strategic partnerships with key technology leaders will gain an important first-mover advantage as they will have access to the latest models and the engineering talent behind them. Finally, our experience has shown that technical skills such as data science have had an amplified business impact when embedded within the investment teams; this will also be the case for AI.

Like the software industry, generative AI can be divided into model layer and application layer. There are already numerous tools being built upon the foundation models such as the recently announced Microsoft Office Co-Pilot for office tasks or Github Copilot for coding.

In the not too distant future, tools that are useful across many companies will be provided by external providers. Tools that are company specific, and can be a source of competitive advantages, will be built in-house. External tools will become standard for doing business across companies. Proprietary tools will be where the competitive advantage lies.

By automating routine and repetitive tasks, AI has the potential to significantly boost productivity, freeing up human workers to focus on more complex and creative tasks that require human ingenuity.

Limitations and challenges

While AI has the potential to revolutionize private market investing, it is important to recognize that there are risks associated with its use.

There are three broad categories of “un-AI-able” tasks that require humans to be “in the loop” to make key decisions.

• Critical thinking
• Conflict resolution
• Broad contextual awareness

These skills are already essential for investors and will be even more so in the future as AI helps with other tasks.

A human centric approach can oversee the AI system and ensure that the output is correct and explainable. Additionally, it is important to recognize the limitations of the technology - while AI can analyse vast amounts of data and identify patterns, it is not a magic solution and can still make mistakes.

Legal, compliance, privacy and security considerations are a pre-requisite, as AI systems must adhere to legal regulations and ethical guidelines. Finally, it is crucial to have governance mechanisms in place to evaluate and approve use-cases, ensuring that AI is being used in a responsible manner.

By carefully considering these risks and implementing appropriate safeguards, private market investors can harness the power of AI while minimizing potential downsides.