The AI way

Applications of artificial intelligence in the asset management industry has become increasingly broad, running the gamut from advanced data analysis to risk management.

Andrew Hendry, Asia chief executive officer of Janus Henderson Investors, shares his thoughts on how cutting-edge AI technologies such as generative AI (GenAI) and large learning models (LLMs) are reshaping the industry landscape.

How has AI evolved over the past decade? What impact has it brought to the asset management industry?

In the last decade, artificial intelligence has undergone significant evolution, driven by advances in deep learning, neural networks, and the introduction of transformative models like bidirectional encoder representations from transformers (BERT) and generative pre-trained transformer (GPF), notably enhancing image, speech recognition, and natural language processing capabilities.

This progress, coupled with increased computational power and the vast availability of data, has made artificial intelligence more accessible, influencing various aspects of daily life and industries from consumer tech to healthcare, where it has improved diagnostics, drug discovery, and personalised treatment.

The era also spotlighted the ethical dimensions of artificial intelligence, sparking debates on privacy, bias, and regulation.

In asset management, artificial intelligence has just begun to be tackled by most global asset managers, though the potential use cases and impact are quite broad.

We see the applications for the asset management industry span almost the entire value chain, from obvious areas like increasing employee productivity through the use of more efficient tools, to the very fundamental, such as the potential for artificial intelligence-driven investment strategies to replace traditional indexing.

Its integration means new skill sets in data science and machine learning are necessary, which changes the makeup of most investment firms’ technology expertise.

Ultimately, what’s most important for us at Janus Henderson and what all of us here endeavour to achieve is positive client outcomes, whether it be in service or investment performance. We believe effective integration and application of artificial intelligence is key to helping us reach this goal.

Please elaborate on the adoption of AI technologies such as GenAI and LLMs in the industry.

No asset manager today can operate efficiently without having a robust technology platform in place. Very soon, successful integration of relevant artificial intelligence technologies will be a similar necessity. Asset managers who fail to adopt and successfully implement these tools are in danger of being left by the wayside.

GenAI and LLMs have the potential to significantly disrupt asset management by enabling advanced data analysis, financial report interpretation, and risk management. They process vast datasets quickly, identifying patterns and insights for informed decision-making.

LLMs, through natural language processing, can efficiently summarise complex financial documents and aid strategy formulation. These technologies not only enhance decision-making but also minimise manual errors. Combined, this should result in a significantly improved client experience.

What are the challenges for asset managers in the course of combining AI with their existing operational systems?

The largest challenges generally faced by asset managers currently trying to integrate artificial intelligence solutions are lack of specific domain expertise, regulatory and privacy concerns, opacity/reliability of models, and importantly, data access issues

For Janus Henderson, artificial intelligence is starting to help our business in large-scale data processing. We consume a number of external data sources, which all need to be normalised, validated and cross-checked with both existing data before being loaded into our systems.

There is noise in all sets of data, which takes time for a human to sort through, identify the source of a discrepancy, and figure out how to correct it. Artificial intelligence-based tooling can not only help identify the errors but also process corrections using reasonable, human-like analysis and cross-validation.

Other solutions can include adopting robust data governance, leveraging cloud computing for infrastructure needs, ensuring artificial intelligence systems comply with global regulations, and developing in-house expertise or partnering with specialists.

Gradual artificial intelligence integration, coupled with comprehensive training and stakeholder engagement, can ease the transition. Addressing ethical concerns and potential biases in artificial intelligence models through regular audits is also crucial.

When did Janus Henderson begin to adopt AI for investment? How does the company strengthen its AI investment capabilities?

Following a planning period, the firm established a data science team in early 2023. One of our strategic initiatives is the formation of what we call our Ignition Hub late in 2023, where we find ways to identify, understand and execute on longer horizon developments that have a potentially disruptive impact on the way we manage our business and clients manage their investments.

It’s still early days, but artificial intelligence is one of the technologies we’re focused on. For example, we have an internal tool available to all employees called Chat JHI where anyone in the firm can go in and utilise GPT-4 from OpenAI in a secure environment within our technology platform and can benefit from the use of Chat GPT today.

We’ve seen some generic use cases today with internal teams using Chat JHI for things like accelerating the process for writing code in Excel or summarising/transcribing meetings, but the potential for how we can harness this to improve performance and efficiencies is so vast. Currently, we have a data science team that is proactively building out specific implementations for longer-term high value use cases.

We’ve also spent a lot of time engaging our employees around these emerging technologies, hosting small group sessions to help educate them, and that’s been a big part of our strategy: educating and then empowering our employees to understand and utilise technology.

We are also beginning to look at the use of artificial intelligence tools in our investment research capabilities. Rather than the outcome of complex analysis that is best done by a human, artificial intelligence tools help our team gather and organise foundational, wide-reaching and time-consuming background data on our research topics, greatly reducing the time spent on the background data gathering and collation, and focusing more of our expertise on analysis and conclusions.

We are also seeing artificial intelligence used in the monitoring/alerting of non-standard system conditions across our various platforms. Often, a singular failure of some infrastructure component like a network switch causes all manner of downstream warnings and alerts and a tremendous amount of noise. That noise makes it difficult to triage what is really going on, to identify and correct the root cause with the minimum amount of disruption.

Artificial intelligence technology sees across all the noise and can very quickly sort, filter, and identify common elements across a large range of detailed alerts, thus rapidly narrowing down the universe of possibilities and focusing our operator time on true root-cause issues.

Most recently, we kicked off a project in May to implement a new artificial intelligence-enabled customer relationship management capability for the salesforce. This artificial intelligence assisted customer relationship management tool will enable our salespeople to aggregate disparate data points that you can pull into one view, effectively streamlining the intel gathering process.

It’s important to note that we don’t see this type of technology as replacement play. While there will be some functions that AI can perform better, overall, we see AI technology as a natural complement and enhancement to the depth, breadth and nuance of human cognition and assessment.

As it evolves, it will take away more of the foundational, manual, and repetitive tasks that investment professionals perform and focus them on the critical differentiators or assessments points, which often require as much art and experience as science and data to make the best predictive decisions.

Further, we expect artificial intelligence tooling to expand the universe of data that an investment professional can see and make sense of, but identifying more and better signals within the noise that should be considered.

What’s your view on the trend of artificial intelligence investment adoption among asset owners in Asia Pacific?

We expect adoption to increase steadily over time as the technology improves and artificial intelligence output becomes more and more reliable. Much like how the widespread adoption and integration of the internet changed the world in the 90s, we think artificial intelligence will eventually change the way business is done in the asset management landscape, not just in Asia but throughout the world.

How should regulators further strengthen existing regulatory provision to address risks related to artificial intelligence?

We think there is less a need for new regulation than for people to understand, for example, the implication of existing data privacy regulation for how they implement artificial intelligence.

That said, addressing artificial intelligence-related risks requires a multi-faceted approach, and we believe a good start would be to address issues around data privacy and security, as there will be massive volumes of data collected, stored and processed.

Artificial intelligence technologies often operate across borders, making international cooperation crucial for effective regulation.

We are tracking what measures are being taken and what is being considered by regulators across the globe with regards to artificial intelligence. The concerns centre around:

• Reliability and potential bias in the data source
• Risks in financial models
• Governance around use of artificial intelligence and consumer protection

It will be important for financial services firms to be able to demonstrate to their clients and regulators that they have robust governance for artificial intelligence, including safe testing and deployment of use cases, and a clear understanding of how any financial models work, and any inherent biases in those models.