Back to blog

Artificial intelligence may change the way companies issue debt

See blog

Readers' comments

The Economist welcomes your views. Please stay on topic and be respectful of other readers. Review our comments policy.


The illusion that Big Data purveys is that somehow, their computers will be more accurate and predictive than humans. However, this is all based on the faulty premise that the underlying algorithm is entirely objective and bias free. Unfortunately, all algorithms are written by humans, and humans decide the consideration and weighting of various risk factors and data points when they create the algorithm. Thus, while an algorithm will certainly be quick and entirely objective within the confines of its own parameters, that is not the same as saying the algorithm will be less biased than a human. Lending programs suffer from this same bias -- they are created by banking firms, and incorporate the same pre-existing biases that banking firms utilize. The same goes for algorithms that purport to predict crime, or terrorism, or illness, or prosperity, or voting interests, etc. Computers are always faster; however, computers are usually just as biased as their programmers.

Peace Love and Understanding

I'm very skeptical that an algorithm at present levels of technology can accurately account for the myriad of variables that go into determining market value, some of those factors being totally irrational human behavior and others certainly un-quantifiable.
The investment banks don't really know either. They are mostly making it up as they go along. But they can never lose here because they control the bridge across the magical money river.


The bigger picture: AI will soon enable everyone to know what the price for anything should be at any given moment. Once everyone has the same knowledge of what a price should be, that's the end of markets.

LexHumana in reply to MathsForFun_1

What a computer thinks the price SHOULD be is never going to be the same as what an individual consumer is willing to pay -- this is why there are markets to begin with, and why they will never cease. A glass of water may objectively be worth ten cents to the computer, but in a group of consumers a very thirsty man might be willing to pay more at that given moment. A computer can calculate cost (objective), but can never calculate value (subjective).

MathsForFun_1 in reply to LexHumana

The thirsty man might be willing to pay more - but if his device is telling him he can get the water for a fair price as easily as he can get the overpriced water, he won't. There will be exceptions, but in general, once everyone has the same knowledge, you no longer have a market - you just have a price.

LexHumana in reply to MathsForFun_1

His device is not going to quench his thirst. If he needs a drink, he won't care that his device is saying "you are being overcharged by 20 cents!". He will say "to hell with it, I'm thirsty, and this is the only drink available". Hence, markets. If the man is willing to say, "I'll wait for a more reasonably priced drink", then clearly he wasn't that thirsty to begin with.

LexHumana in reply to MathsForFun_1

You can have multiple vendors, yet the consumer may purchase what is most convenient and readily available. Time and convenience have a value as well, and it is subjective to each consumer. You see this in the vending of gasoline, for example -- there is a robust market (stations everywhere), and ample opportunities for price arbitrage (apps and publications showing the price of gas at nearby locations). For those interested in squeezing out a few cents per gallon, they can continue to drive to a cheaper location. For those who are in a hurry, or their tank is on "E", they will fuel up at a more expensive locale so long as it is the closest. The mere fact that the market can figure out what a more "fair" price might be is not the sole driver of what the market is willing to pay, hence the existence of price differentials. AI is not going to change that one whit.