Artificial intelligence can enhance chaos in the stock market, the Bank of England warns

The stock market is already an unexpected place, and now the Bank of England to caution The adoption of obstetric intelligence in the financial markets can result in a mono culture and the amplification of more stock movements. A report issued by the bank’s financial policy committee, which argued that independent robots may learn that fluctuations can be profitable to companies and take deliberate measures to swing in the market.

Basically, the bank is concerned that the phrase “purchase of DIP” may be adopted by models in necknotherapy and that events such as 2010 are notorious “for the year 2010”FlashIt may become more common. With a small number of foundational models that dominate the area of ​​artificial intelligence, especially those in Openai and anthropology, companies can converge in similar investment strategies and create herd behavior.

But more than just following similar strategies, models work on the reward system – when they are trained using a technology called reinforcement learning with human comments, models learn how to produce answers that receive positive notes. This has led to a strange behavior, including models that produce fake information that they know will continue to review. When the models are directed not to compensate the information, it was found that it will work Take steps to hide their behavior.

Fear is that models can understand that their goal is to achieve a profit for investors and do so through unethical means. Artificial intelligence models, after all, are not a human being and do not understand what is right in essence for error.

“For example, models may learn that stress events increase their chance to make profit and thus take effective measures to increase the possibility of such events,” the report also reads from the Capital Policy Committee.

High -frequency algorithm trading is already common in Wall Street, which led to sudden and unexpected inventory movements. In recent days, the S&P 500 has increased by more than 7 % before it declined after a post on social media Comments bad interpretation By the Trump administration indicates that it will stop the customs tariff (which it seems It is already happening nowAfter a previous denial). It is not difficult to imagine Chatbot like X’s Grok, which deals with this information and making deals based on, causing great losses for some.

In general, artificial intelligence models can offer a lot of unexpected behavior before human managers have time to control. Models are mainly black boxes, and it may be difficult to understand their options and behavior. Many have noticed that the introduction of the Apple to the Model Organization in its products is unlimited, because the company was unable to control technology outputs, which leads to unsatisfactory experiences. For this reason, there is concern about the use of artificial intelligence in other areas such as health care Where the cost of errors is high. At least when a person is in control, there is someone to be held accountable. If the artificial intelligence model manipulates the stock market and the trading company managers do not understand how the model works, can they take responsibility for organizational violations such as manipulation of shares?

Certainly there is a variety of artificial intelligence models that act differently, so it is not a warranty that there is a sudden breakdown of stocks due to one model suggestions. Artificial intelligence can be used to simplify administrative work, such as writing emails. But in fields with low tolerance of error, the use of artificial intelligence on a large scale may lead to some bad problems.