• Global Research
    • Data science
    • Disruptive technology
    • General Research Insights

Can AI be your Teammate?

  • Article

How artificial intelligence can help analysts

  • In our latest experiment, we explore how AI could help analysts…
  • …by extracting pricing data from management commentary on earnings calls
  • AI can offer human analysts accurate and speedy support, but there are inherent limits to the role it can perform

Complementing, not competing

Modern AI models can perform well at a wide array of tasks. This has fuelled hopes of significant AI-driven productivity improvements, as well as fears of its impact on job markets. We believe that worries of widespread job losses are overdone – in reality, AI works best when it augments human experts, rather than replacing them.

Combing through the haystack

Having previously examined how AI could support data analysts and economists, our latest experiment explores how effectively AI could support analysts by combing through lengthy transcripts of insurance earnings calls to identify pricing information. This data is key to understanding broader sector dynamics, but can sometimes seem like the needle in the proverbial haystack.

Speedy, accurate…

Our results suggest that AI performs this task with a decent level of accuracy, broadly similar to a diligent junior teammate. Yet, the AI is far faster, taking a fraction of the time a human would need. It doesn’t get bored, and it doesn’t get distracted. But as with our previous experiments, we encountered some challenges and inherent limitations in what AI can reasonably be expected to deliver.

[the AI tool] streamlines the process and gives us more time to focus on higher value activities.

…but not perfect

AI needs to be trained on how to approach the task correctly. It can struggle with mathematical calculations. It can occasionally be too creative in its responses, leading to the dreaded phenomenon of ‘AI hallucination’, potentially undermining trust. And fundamentally, AI does not have a consistent outlook: if you ask the exact same question three times, it may give three wholly different responses.

Understanding these limitations and characteristics is key to calibrating expectations of how best to employ AI, in our view. As an aid and complement to a human analyst, its potential value is clear.

HSBC Global Investment Summit

The second HSBC Global Investment Summit took place on the 25 to 27 March 2025 in Hong Kong, bringing together over 4,000 delegates to discuss the global trends and topics shaping the global economy.

Need help?

For more information, please contact your HSBC representative.