While people have been excited about ChatGPT, fund managers were already investing in artificial intelligence (AI) and implementing it in their industry, but opinions vary as to the extent that this technology will permeate their jobs.
Below, three managers discuss their approach to AI and suggest a picks-and-shovel approach for those looking to invest in the area.
8AM Global’s Merricks: “AI is hastening my own demise”
Andy Merricks, co-manager of the IDAD Future Wealth fund, was surprised it has taken so long for AI to gain public attention.
“It’s been a slow burner, but the world appears to be waking up to its potential or threats,” he said.
“As a fund manager it could be argued that investing in AI is hastening my own demise, but ignoring an investment proposition wouldn’t be doing myself or investors any favours. So I may as well win the price of a consolatory pint to soften the seemingly inevitable consequence of losing.”
The top five professions most under threat from AI include fund management, journalism, legal and accountancy, the manager recalled from a lecture he attended.
“The inevitable is coming down the track, particularly for those who construct portfolios using the same restrictive parameters as most other portfolio managers who narrow down the pool of available funds for them to use to a puddle, leading to them all fishing in that same puddle.
“Outcomes are consequently quite similar yet this seems to satisfy them that they are not that different from anyone else. Why pay inflated salaries in these instances when AI will probably be able to do just as well but without the need to holiday or sick pay?,” he concluded.
Sanlam’s Ford: “We see AI as a complement to human endeavour”
Chris Ford, head of growth equities at Sanlam, said AI’s way into asset management will be through the performance of dull, repetitive tasks, which humans tend not to be particularly good but machines excel at.
“In the front office, AI can help to automate some of the ‘insight’ parts of the investment process. In the middle and back office, it can improve efficiency in fund pricing and administration and free up human teams to concentrate on more complex or interesting tasks. It can also be used to improve risk monitoring and oversight,” he said.
“Within financial services more generally, AI has also been deployed to great effect in defeating fraud – again this is because it is very good at spotting anything that is out of kilter with the norm.”
The managers of the Sanlam Global Artificial Intelligence fund worked with a partner to develop an AI tool that utilises several different models (natural language processing, optical character recognition and bidirectional auto-regressive transformer) to screen qualitative data in multiple languages.
“The tool has been a key element of the investment process since the fund launched in 2017. It searches company documents, stock market notifications and other documents that are ‘on the record’ for the markers of AI that we define and allows the team to identify companies that are significantly engaging with AI in a way that affects their economic prospects,” said Ford.
“AI can also complete some of the more commoditised areas of stock research, for example generative AI models can be used to summarise what a company does or doesn’t do in terms of its general activities.”
The actual stock selection within the strategy is still, however, a human-led process.
“We don’t expect that to change. The pace of change of AI is so rapid in some areas that we constantly review our search criteria to ensure relevance,” he said. “In most cases, we see AI as a complement to human endeavour, rather than a direct replacement for it.”
Baillie Gifford’s James: “Most companies will tell us that AI is a good thing for them, but they won’t all be correct”
Ben James, investment specialist for US equities strategies at Baillie Gifford, equated the breakthroughs in AI to the revolutions of the personal computer, the internet and smartphones.
“AI and machine learning are penetrating multiple industries, from commerce and healthcare to finance and transport and will likely transform many parts of the economy, but it would be foolhardy to make specific predictions. Any company not leveraging advances in software, machine learning, and AI, however, will likely experience significant headwinds to growth,” he said.
“Most companies will likely tell us that AI is a good thing for them, but they won’t all be correct. At the same time, it will substantially increase the value of the service other companies can provide. Patience and a long-term mindset are key in deciphering signals from the noise amidst all the AI hype.”
Baillie Gifford has been experimenting with and thinking about AI for a long time, with several analysts at the firm, including in the US equity team, using a tool called AlphaSense, which James described as a crossover between Bloomberg and Google.
“It uses AI to gather content to help analysts get data on companies and markets. For example, if an analyst wants to see what a company has said historically about its plans for cutting carbon emissions, then they can use AlphaSense to scrape previous earnings calls transcripts and pull it all together in one place, rather than spending hours going through PDFs themselves,” he said.
“We are also experimenting with OpenAI’s ChatGPT across the firm, including on the investment floor. At this stage, we’re using it for things such as help in report writing (editing and challenge).”
The picks and shovels of AI
From an investment perspective, AI systems will require a lot of silicon, which will benefit companies like the much-hyped NVIDIA, said James.
“OpenAI has suggested that the computing power needed to run the latest models doubles every 14 weeks. Our holding NVIDIA is a key supplier and enjoys formidable advantages, as the chip technology it has built over decades for computer games has proven ideally suited for AI computation. Around 90% of generative AI programs are currently trained using NVIDIA chips. The market is making a lot of noise now about NVIDIA as a way to invest in AI, so one of our challenges is assessing how much of NVIDIA’s future growth from AI is already priced in.”
Going beyond NVIDIA to make some ‘picks and shovels’ investments in “the great AI gold rush” (meaning investing in the enablers of a technology rather than the technology itself), Baillie Gifford American is exposed to a diverse range of companies across the AI value chain, such as Amazon Web Services, on whose cloud much of AI computing is done, and cloud services and enterprise solutions that operate on top of the cloud, such as Cloudflare and Snowflake, said James.
“Some industry solutions already use AI such as Recursion Pharmaceuticals (drug discovery). Consumer-facing services such as Duolingo (the language learning app) and the digital marketing and metaverse companies we hold, such as Roblox, are also experimenting with AI in their products,” he explained.
“It may surprise some to learn that Tesla is currently one of the largest AI companies in the world. It rolled out initial access to its full self-driving software in the US this year. The system’s capability is already impressive, but the pace of improvement will be most important over time.”