Q1 What is the influence of AI in finance?
To start with, AI is not the revolution everyone is talking about, it is developing slowly. AI is based on historical data, of which we can process more and more at an increasing rate. As a result, computers can see connections that we didn't see before. This allows us to describe much better what is happening in the financial markets. A next step could be for AI to predict market developments and prices through pattern recognition. That would of course be the ideal. The problem with financial markets, however, is that there is always a lot of noise in the data. There are few reliable signals. I wouldn't let AI make investment decisions. At most, it can help, but in the end the investor has to do it himself.
Q2 So what is the value?
Big data allows us to collect and analyze not only prices, but also news reports and other background information via automated LLM models (similar to ChatGPT). Increasingly powerful computers and more sophisticated algorithms can draw increasingly specific conclusions from information in the field of price developments, inflation, interest rates, risks, you name it. This makes it possible to make interconnections between, for example, geopolitics, inflation and price movements. In any case, this allows for more well-founded investment decisions. Again, this does not offer any certainty, because no matter how sophisticated pattern recognition is, history never repeats itself exactly. Events that occur for the first time, such as corona, cannot be predicted by AI either. The markets are also far too complicated for that.
Q3 How do asset managers and financial institutions use AI?
With all the detailed information, a financial analyst can help many more clients in a better way than before, because AI can also generate relevant financial advice quickly. AI is also having an increasing impact on the back office: opening accounts, transferring money, everything is more efficient and smoother.
Q4 Do you expect an investor to ever leave their decisions to a computer?
At least not in the near future. Just as a medical professional does not let the scanner draw the conclusions of an MRI scan alone, an investor will not do so with his assets. Important decisions with a major impact are not left to a computer that you don't know exactly what it bases its decisions on. In the distant future that may be possible, who knows.
Q5 How do you see the future in this area?
We will trust machines more than we do now, and have a better understanding of the situations in which AI models do or do not work. There will be exponentially more computing power and more data, making more refined models possible. As a result, risks can be better managed, money laundering and other fraud will be more easily recognized and combated. Investment products will be assessed faster and better for their risks and opportunities, and can be linked to the wishes and risk tolerance of individual investors in no time.
The influence of AI on market behavior is still difficult to predict. Because information is processed faster and faster, the question is whether there will be no overreactions in stressful situations, or whether AI will have a stabilizing effect in turbulent times. The latter is supported by the fact that more and more different AI models are being developed, so that price reactions will no longer all go in the same direction automatically.
Medical Analogies for Doctor Audience
Connecting finance and AI concepts to medical practice for 35,000 Dutch medical specialists
"Just as a medical professional does not let the scanner draw the conclusions of an MRI scan alone, an investor will not do so with his assets. Important decisions with a major impact are not left to a computer that you don't know exactly what it bases its decisions on."
| Finance/AI Concept | Medical Analogy | Explanation |
|---|---|---|
| AI in investing | AI in radiology | Assists but doesn't replace the specialist |
| Model risk | Drug interaction risk | Models can interact unexpectedly |
| Black box problem | Drug that works but mechanism unknown | We use it, but can't fully explain why |
| Overfitting | Treating symptom, not disease | Model fits noise, misses real pattern |
| Human+AI collaboration | Surgeon with robotic assistance | Technology extends capability, human directs |
| Risk management AI | Vital signs monitoring | Continuous, alerts to anomalies |
| Flash crash | Anaphylactic shock | Sudden, systemic, requires fast intervention |
| Market volatility | Chronic vs. acute conditions | Different management approaches |
| Portfolio diversification | Multi-drug regimen | Not all eggs in one basket |
| AI screening tools | Screening tests | High sensitivity, may need human confirmation |
| Robo-advisors | Telemedicine apps | Good for routine, limited for complex cases |
| Algorithmic herding | Antibiotic resistance | When everyone uses same approach, it stops working |
Key Insight: Dr. Osterrieder's MRI analogy resonates directly with this audience: AI is a powerful diagnostic and analytical tool, but the specialist must always interpret and decide.
Dr. Jörg Osterrieder
Associate Professor of Finance & AI, University of Twente | Professor of Finance, Bern Business School
COST Action Chair
400 researchers, 51 countries
Chair of Europe's largest research network on Fintech and AI in Finance (CA19130), Horizon Europe funded
Marie Curie Network
EUR 3.8M, 20+ institutions
Coordinator of EU-funded Industrial Doctoral Network on Digital Finance with 100+ researchers
Industry Experience
Goldman Sachs, Credit Suisse, MAN AHL, BCG
Quantitative trading, algorithmic strategies, and strategic advisory in financial services
Latest Research
"Generative AI for Finance Professionals" (2024)
12,400+ views on "Machine Learning for Asset Management" - comprehensive 4-part resource