Some have always suspected: The benefits of systems like ChatGPT & Co. are not worth the noise made about Artificial Intelligence (AI). At least they now have confirmation. Jim Covello, Head of Global Equity Research at Goldman Sachs, said in an interview for the US financial institution: «People generally overestimate what technology can do today.»

Banks exceed expectations with assessments

Artificial intelligence is one of the big topics in the industry. No financial institution is without a study or a guide on it.

UBS published a «framework to assess and seize AI investment opportunities» last June. The investment specialist at Zürcher Kantonalbank explained in a blog post last summer what challenges and investment opportunities AI offers. And for the Swiss Bankers Association, it is clear: «Artificial intelligence is considered one of the most promising technologies of the digital transformation of financial institutions.»

The 1 trillion dollar question

All this is certainly not wrong. But according to Covello, it is essential to critically question the hustle and bustle. «My main concern is that the significant costs of developing and operating AI technology mean that AI applications must solve extremely complex and important problems for companies to achieve an adequate return on investment (ROI),» he says.

Building AI infrastructure is estimated to cost over 1 trillion dollar in the coming years, including spending on data centers, utilities, and applications. Covello: «Therefore, the crucial question is: What 1 trillion dollar problem will AI solve? Replacing low-wage jobs with extremely expensive technology is fundamentally contrary to the previous technology transitions I've followed in my thirty years in the tech industry.»

Hesitant use in Switzerland

In Switzerland, AI is mainly used in customer contact at financial companies today, as the Lucerne University of Applied Sciences and Arts recently showed in a report. However, the study also shows that most deployed bots currently work on rule-based systems, meaning they only respond to predefined questions.

These bots are not able to automatically transfer the conversation to a human employee if a question cannot be answered. Customers are then forced to seek contact independently through another channel. In contrast, intelligent AI-based bots are more flexible and can seamlessly transfer conversations to human employees, significantly improving customer service.

Not comparable to the early days of the internet

AI is often compared to the early days of the internet today. Jim Covello of Goldman Sachs does not allow this comparison: «Even in its early days, the internet was a low-cost technology that enabled e-commerce to replace expensive existing solutions. Amazon could sell books cheaper than Barnes & Noble because it did not have to maintain expensive brick-and-mortar locations,» he argues. The tech world is too complacent in assuming that AI costs will significantly decrease over time.

He also questions the possibilities attributed to AI. «I doubt that technology will ever achieve the cognitive reasoning required to significantly expand or replace human interactions.»

Not for the first time: High costs, little benefit

The Goldman Sachs specialist can very well judge this. He was a semiconductor analyst when smartphones were first introduced. «I witnessed hundreds of presentations in the early 2000s about the future of the smartphone and its functionality, much of which has come true as the industry expected,» he says.

It wouldn't be the first time that hype has triggered investments that ultimately do not pay off. Virtual reality or the metaverse are two recent examples. According to Covello, it hasn't come to that yet, even though Nvidia shares have recently taken a hit: «As long as corporate profits remain robust, these experiments will continue. Therefore, I do not expect companies to cut back on spending on AI infrastructure and strategies until we enter a more challenging phase of the economic cycle, which we do not expect anytime soon.»