Although Artificial Intelligence has been a hot topic in the City of London since the launch of ChatGPT, it remains quite opaque who is actually already using generative AI applications and to what extent. However a look at the traditionally conservative asset management sector shows that preparations are well underway.

Almost half of the industry representatives surveyed by Boston Consulting Group stated that their company has already prioritised more than ten application possibilities. In 27% of cases, at least one application is already in use. Another 39% plan to be at this stage within six months. More than two thirds expect significant, if not transformative, impacts of the technology on their business.

Asset managers, facing rising costs and falling margins, hope that the new technology will ease the strain. However, implementation involves sizeable investments. „Everyone wants to invest in AI, but where is the money going to come from?“, asks Trevor Schulze, Chief Information Officer of the AI platform provider Alteryx. „It's almost as if you can see it, but you can't touch it.“

The IT industry veteran, who previously served as Global CIO of US semiconductor manufacturer Micron, deals with companies outside the financial sector as well. Some companies struggle to balance cash flow, while others have laid off employees to finance technology investments.

„Building models from scratch is very expensive“, comments AI consultant Alex Dollery, who previously served as Director of AI & ML Services, Product Management at the London Stock Exchange Group. „You can do it, but it takes a long time to update them.“ On the other hand, small models offer many advantages.

Asset managers affiliated with major banks have an advantage over independent providers. They can leverage an IT infrastructure that they might not otherwise be able to afford.

„The introduction of AI can require significant investments for financial firms“, says Euan Munro, CEO of Newton (BNY Investments). BNY was the first major bank to have an Nvidia DGX Superpod, a data center solution providing accelerated infrastructure for AI and high-performance computing.

„We use Large Language Models (LLMs) and proprietary Newton research to help answer questions more quickly, such as: Which companies can give us exposure to certain markets? Which companies' businesses could be influenced by elections?“, explains Munro. This helps analysts and researchers identify opportunities and sources of risk. „We use AI to become more perceptive“, notes Munro. „We absolutely do not believe that machines will replace people.“

Many asset managers use the new technology to summarise the daily flood of broker emails. It is also popular as a „copilot“ and idea generator, especially among younger employees. In some places, AI applications follow meetings, highlighting points that might otherwise be overlooked.

„I don't think AI is being used at an advanced level anywhere in the fund industry yet“, states portfolio manager Alan Edington of Walter Scott (BNY Investments). He does have a tool that aggregates research and expert opinions. „It does this very well and is very helpful“, says Edington. „It saves a lot of time. But we are far from it being able to tell us what the right investment decision is.“

Jeffrey Lin, fund manager of the M&G Global AI Themes Fund, uses AI to evaluate the daily flood of information. He relies on a traditional bottom-up model for his investment decisions. „That requires a lot of judgment“, notes Lin. „AI is not capable of that at this point.“

Banks use AI to prevent fraud, such as filtering out unusual movements in transaction banking. Some financial sector employees use AI to get suggestions for improving longer emails. Especially when working from home, a lot of content might be copied back and forth between private and work applications, posing new risks. „A mistake in assessing a customer's creditworthiness and it's over“, says the head of the corporate banking division of an international major bank.

In the hedge fund industry, there is comparatively more reticence, which might surprise readers of Robert Harris' thriller The Fear Index. This is due to increasing regulation of the business, and investors' desire for more transparency.

Even when a firm wants to use „alternative data“ such as weather data for a strategy, it must be approved by the investment committee, says Grace Lo, Chief Risk Officer at the quant fund Campbell & Co. She tries to use AI to filter ideas from sell-side publications.

ChatGPT delivers „confidently wrong answers“, says Lo. There is also a risk of intellectual property leaking through the use of generative AI. „For well-defined, repetitive tasks, it works well“, adds Lo. The HR department uses ChatGPT to ask better questions in job interviews.

Generative AI offers enormous potential for streamlining workflows. Last year, an Accenture study concluded that 54% of jobs in banks have high automation potential. For insurers, it was 48%.

The optimistic view is that those affected will have more time for interesting work in the future, because AI will take over boring and repetitive tasks. „We will not run out of work“, says Dollery. In the end, it's an ecosystem that has to function for everyone. If everyone lost their jobs, no one could afford a BMW anymore.

At Bank of America, the virtual assistant „Erica“ has been in use since 2018. Since then, it has answered more than 800 million inquiries from 42 million customers. The bank adopted the underlying technology for its corporate client platform Cashpro last fall, helping over 40,000 business clients manage their treasury activities.

Unfortunately, it is likely that junior analysts, who previously listened to company earnings conference calls and are now replaced by AI, will not have more time for in-depth research, says Kier Boley, co-head and Chief Investment Officer Alternative Investment Solutions at Union Bancaire Privée. It is more likely that employers will do away with this cost level.