Deutsche Bank research reveals what AI is still bad at ...

Despite its rapid rise, generative AI is not without its limitations. A recent report by Deutsche Bank Research underscored that while the technology excels in certain areas, it still struggles with tasks like mathematical calculations .

While generative AI has proven valuable for tasks such as summarization, translation, and creative content generation, it remains challenged by reasoning, abstract concept learning, and understanding the world. The report emphasized, "Generative AI is certainly flawed... while it's surprisingly good at some activities, it's surprisingly bad at others, such as making mathematical calculations."

Biggest pain points of AI
One significant issue is the tendency of generative AI systems to produce hallucinations, or inaccurate information. These systems can also introduce bias or irrelevance into their outputs. Despite ongoing efforts, these problems persist.

The report further highlighted that while AI's potential for productivity gains is promising, real-world applications may not always deliver. Highly regulated industries like finance and healthcare, where errors can have severe consequences, have been cautious about adopting generative AI.

While generative AI has shown potential in unexpected areas, such as generating novel research ideas and creating game engines, its full potential will only be realized with continued development and careful implementation.