APG: How it strikes the man-machine balance with AI
APG asset management has a firm belief in the “man-machine” balance, a concept the fund introduced back in 2017 when it first began to experiment with artificial intelligence (AI), according to Peter Strikwerda, global head of digitalisation & innovation, at the $615 billion pension investor.
“We use machines to augment our intelligence but never to fully automate decisions or execute them without human oversight. There are always people involved for any type of decision in our investment portfolio,” Strikwerda told AsianInvestor.
Human decision-making remains paramount, but AI-driven modules are currently being leveraged within the fund’s investment strategy.
APG
“We have something we call ‘real-time trade analytics’, which advises our traders on the best strategy to place a particular trade – for example if we want to put $50 million into Company X. The AI-driven module provides guidance on how to do it in specific portions at the best times of the day and in the right venues,” he said.
APG’s most famous AI model is called “Samuel”, a digital portfolio manager used to make information and analytics instantly accessible to the fund’s alternatives investment team.
“It’s a big deal, because in private markets, information is not as well organised as in capital markets, so using generative AI and large language models, someone with no programming experience can use prompts to access the right data,” said Strikwerda.
The next step for the firm will be creating a predictive AI model, which could potentially help determine which alternative asset would reap the greatest return on investment.
“That's something we're experimenting with currently,” he said.
THINK BIG, ACT SMALL
Since initiating the experiments, APG has adopted a structured approach to assess the potential value of AI throughout its entire value chain. Part of its process includes keeping a regularly updated heat-map.
“This basically is a bit of a top-down driven approach to pinpoint areas where we think the impact of AI, and specifically generative AI, will be the biggest,” said Strikwerda.
“We will choose a few of those areas with the biggest impact and an acceptable risk assessment. Then we typically think big and act small. Which means we will start with an experiment, for example, or by creating an infrastructure so that colleagues can experiment and gather proof points, to determine whether it makes sense or not.”
For heavily scrutinised pension investors like APG, managing the balance between opportunity and risk is extremely important when it comes to AI.
“I'm proud that our board has taken a proactive stance on adopting AI,” said Strikwerda. “We're engaging with regulators to address the dilemmas in risk management and striving to be at the forefront of AI implementation in investments,”
AUGMENTED INTELLIGENCE
Asset manager T. Rowe Price has also been building capabilities in data science, machine learning, and predictive analytics since 2017 in support of its investment associates, according to Jordan Vinarub, head of the firm’s New York City Technology Development Center.
T.Rowe Price
“Through access to alternative data, solutions in natural language processing, and predictive models, we have been able to empower our investment decision makers with data driven insights,” Vinarub told AsianInvestor.
Like APG, the asset manager has adopted a strategy of “intelligent augmentation” as opposed to automating decision-making through AI.
“We bring the power of data and insights to the human decision maker in their existing process,” Vinarub said. “This approach has enabled us to evolve our capabilities in a thoughtful way, leveraging the capabilities of AI to help shift the human decision maker to higher value tasks within the investment process.”
As generative artificial intelligence (GAI) continues to rapidly evolve, constant collaboration and assessment of new offerings is essential to understanding how best to incorporate the technology into the business process, he said.
“We see GAI supporting the 3 C's: consumption, characterisation, and creation. As the amount of information flooding our research analysts grows, GAI can serve to help by quickly consuming text content, finding signal within the noise, and facilitating the creation of new content in notes, emails, and presentations,” Vinarub elaborated.
These capabilities promise to provide a productivity lift to business processes, as well as to the firm’s investment process overall.
“The firms that navigate these changes and learn how to integrate the capabilities into their business will be the ones best positioned to evolve and grow,” Vinarub said.