Don’t Bet on AI Traders Just Yet
Traders fearing that they eventually will be replaced by HAL 9000 or its artificially intelligent kin can relax for a bit, say experts.
“I believe that everything a human being can do can be programmed,” said Stephen Denier, CEO of industry consultancy Biji Advisors. “It is just a matter of time. However, those programs also will include the predictable biases and mistakes that we make.”
With his background in cognitive sciences as well as a history on Wall Street, Denier is not convinced that someone can wind up an AI, point it in the right direction, and let it learn everything it needs to know.
The AI is dependent on having its programmers set it off in the right direction, he said. “This means that they have already asked and answered all the right questions to figure out where the AI needs to go.”
Denier doubts that it truly is possible for programmers and model makers to do that.”There are very few people in this business, systematic trader or otherwise, who generate positive returns consistently,” he noted.
Until AI developers can overcome their cognitive biases, such as which data sets they select for AI and how the AI should weight the data in its analysis, the AIs will incorporate their developers’ biases, according to Denier.
“What is happening now is that AI developers are trying to replicate human beings, complete with their limitations, but without emotions,” he said. “Emotions aren’t the only biases in our decision-making process. There are many other perspectives and emotions are just one of them.”
For AI and machine learning applications to progress beyond human performance, they’ll need to think outside of the infamous ‘box’, which Denier defines as the sum of an individual’s experience and knowledge.
“People cannot think outside their on boxes, but they can think outside of someone else’s box,” he explained. “AIs do not realize that they only have access to a certain set of data. They believe that all the data to which they have access is the entire universe.”
Denier cites the role of the U.S. Federal Reserve to illustrate his point.
“If you understand that the target audience of the Fed fundamentally has changed as a result of wealth disparity then you understand that whatever the Fed does is a reaction to that function and not necessarily what the market has seen in the past,” he said. “How would an AI know this? The AI would not know the right questions to ask beyond the fence that its programmers developed for it.”
More on Artificial Intelligence:
Data extraction and integration is the second stage of a digitization process.
Increased electronification has created useable and accessible real-time and historic trade data.
The fintech uses data so institutions can assess the environment impact of their portfolios.
Working with Riskfuel will reduce reliance on slow and expensive financial models.
The platform uses AI to match experts and users and digitize the research process.