The Stanford study on artificial intelligence (AI) and white-collar jobs has caused quite a stir. In an interview, Client Portfolio Manager Oleg Schantorenko explains why productivity boosts are still a long way off, why office workers in particular need to tremble - and why Europe has fallen behind in the global AI race. The interview appeared in the magazine "Trends in Asset Management".
TiAM FundResearch: Mr. Schantorenko, the Stanford study on the impact of generative artificial intelligence (AI) on the labour market has attracted a lot of attention. Did you expect the results?
Oleg Schantorenko: Yes, basically. We had already discussed the topic internally in the summer and the Stanford study was published shortly afterwards. Of course, the widespread media coverage took away some of the element of surprise from our findings, but we feel that our assessments have been confirmed.
White-collar jobs seem to be particularly affected. What developments do you see there?
The Stanford study shows: Computer jobs in particular are surprisingly easy to automate - much easier than physical work. Developers, accountants, assistants and consultants are particularly affected. In traditional consulting firms, we are already seeing a decline in junior positions: Teams of 15 people used to be deployed, today perhaps two to three employees still do the same work together with AI tools.
However, many people previously assumed that blue-collar jobs were more at risk from automation. Is that a misjudgement?
Yes, at least in the short term. Surprisingly, cab drivers and other drivers are considered relatively protected professions in the studies, while business services such as research, scheduling or presentation preparation are most affected by AI, according to the Stanford study. White-collar jobs in professional business services - a sector with over 23 million employees in the USA - are already showing a clear decline, while no slumps are visible in construction or the hotel industry, for example.
What role do managers play in this development?
A central one. Everything that has to do with decision-making is difficult to automate. AI can support, analyze and prepare - but the responsibility for a decision remains with the human being. Decisions have consequences, which is precisely why managerial positions are much better protected.
And the employees? Should we expect mass redundancies?
No, at least not across the board. It is more likely that companies will no longer create new positions to the same extent as they grow, but will use AI to make existing teams more efficient. This is more a case of a "changed growth path" than acute waves of redundancies. This can be a problem for career starters.
Many observers expected an immediate boost in productivity after the "ChatGPT moment". So far, however, the data shows otherwise. How do you explain that?
That's right. Productivity in the US - measured as output per hour - is well below the long-term average. Instead of immediate efficiency gains, we have so far mainly seen investments: in data centers, in research, in internal project teams. This is a kind of upfront investment. The positive case would be that these investments are reflected in rising productivity in a few years' time. The negative case is that the models simply do not yet find the problems for which they are supposed to be the solution.
Do you see the current development of AI as a milestone, comparable to the internet or smartphones?
Yes, definitely in the long term. But as humans, we tend to overestimate in the short term and underestimate in the long term. The actual penetration into the economy and society will take time - perhaps a decade. A good example is video streaming: initially, people were skeptical as to whether customers would really pay for it. Today, it is standard in most households.
But so far, the fields of application for AI have remained rather limited, haven't they?
Exactly. Many AI applications are currently more of a "little helper" - for example in text summaries or presentations. Truly groundbreaking applications for the masses are still lacking. Monetization is also a problem: investments amount to hundreds of billions of dollars, while revenue from subscriptions or business solutions has so far been disproportionate.
What does this mean for companies and industries?
In the tech sector, hardware providers in particular are currently benefiting because a huge amount of infrastructure is required. Software companies can enrich their products with AI - some will win, others will lose. In traditional industries, we are currently still very hesitant about the use of AI. There, the topic is more likely to find its way into existing processes by increasing efficiency.
Europe has so far only played a minor role in global AI development. Why is that?
Europe focused on regulation early on with the AI Act - tactically unwise in my view. While capital, innovation and talent come together in the USA, Europe lacks venture capital and attractive start-up ecosystems. Many bright minds are emigrating. In fact, Europe today is more of a consumer than an innovator.
Could this still change, or has the train already left the station?
To be honest, the big platforms are no longer being created here. Of course there are interesting start-ups in Berlin, Barcelona or London, but after two and a half years of the ChatGPT era, we don't see a European success story. I expect niche solutions at best, while the economies of scale lie with US and Asian players.
Where does the USA stand compared to Asia?
The USA is represented at practically every stage of the value chain - from chip design to foundries and software. Asia, especially Taiwan, plays a key role in production. Europe only has a selective presence. The USA's real weakness is rare earths, but otherwise it is almost self-sufficient.
You mentioned the investment perspective: which sectors are particularly benefiting from AI?
Semiconductors and AI are currently the few global growth segments with 20 percent or more growth per year. As asset managers, we can't avoid these markets, even if the dollar risk has to be managed. A second growth area is armaments, although this is currently not investable for us for sustainability reasons.
How does DJE position itself in relation to AI?
We have a clear AI policy. This means that AI is used to reduce repetitive tasks and increase productivity. At the same time, we pay strict attention to data protection - sensitive client data must not be uploaded to open models. In research, we are already using AI to supplement traditional sources such as brokers or Google.
Many people also see risks - such as a flood of information that no one can process. Do you share this concern?
Yes, that is a real problem. AI can inflate small pieces of information, seemingly creating more complexity than is actually there. The bloated document then ends up in another tool for summarization. This creates an inefficient cycle. It remains crucial that we use AI sensibly - not to artificially enlarge content, but to create real efficiency.
Finally, your market outlook: How is DJE currently positioned in equities, bonds and gold?
Equities remain our preferred asset class because they represent real assets and offer a certain degree of inflation protection. Corporate bonds are currently quite highly valued, which is why issuer selection plays a major role. Government bonds appear attractive at first glance, but are suffering from high budget deficits and rising interest rates. Gold is the clear beneficiary of the current situation - as a safe haven and inflation hedge.
Legal information
Marketing advertisement: All information published here is for your information only and does not constitute investment advice or any other recommendation. The statements contained in this document reflect the current assessment of DJE Kapital AG. These may change at any time without prior notice. All statements made have been made with care in accordance with the state of knowledge at the time of preparation. However, no guarantee and no liability can be assumed for the correctness and completeness.