Is Artificial Intelligence like a typical economist?

Diane Coyle wrote a piece on the Project Syndicate website discussing that computers are designed to think like economists. Artificial intelligence (AI) is a faultless version of homo economicus as it is a rationally calculating, logically consistent, ends-orientated agent capable of achieving its desired outcomes with finite computational resources. They are perceived as much more effective than a human in achieving the maximum amount of utility for an individual. Coyle does go onto say that economists today cannot offer a measure of actual utility.

Jeremy Bentham’s famous formulation of utilitarianism is known as the “greatest-happiness principle”. It holds that one must always act so as to produce the greatest aggregate happiness among all sentient beings, within reason. John Stuart Mill’s method of determining the best utility is that a moral agent, when given the choice between two or more actions, ought to choose the action that contributes most to (maximises) the total happiness in the world. However this assumption can produce some unease.

  • Most of those designing algorithms are utilitarians who believe that if a ‘good’ is known, then it can be maximised. Therefore how much thought is there about possible societal impacts of algorithms as they are designed to optimise efficiency and profitability.
  • Algorithms are created using current and future data that is full of bias. The result could be the institutionalisation of biased and damaging decisions with the excuse of, to quote ‘Little Britain’, ‘the computer says no’. see video below.
  • Algorithms make it easy for consumers to decide things and it acts as a short-cut (heuristic). Therefore we become a slave to the algorithm rather than taking more ownership of our thinking /reasoning. Those who  control of the algorithm have an unfair position.

There is no doubt in certain aspects of society AI is extremely useful and can cut down bureaucracy and lead to improved efficiency in everyday life. The real issue extends beyond the use of algorithmic decision-making in corporate and political governance, and strikes at the ethical foundations of our societies. As Coyle points out we need to engage in self-reflection and decide if we really want to encode current social arrangements into the future.

High-Frequency Trading (HFT) – Speed Kills

Flash CrashJames Surowiecki (writer in the New Yorker) wrote a very informative review (in the New York Review of Books) of Michael Lewis’ book ‘Flash Boys’ about the rise of high-frequency trading (HFT) on Wall Street. As the name suggests, high-frequency traders buy and sell in large volumes and at an extraordinary fast pace, trading thousands of times a second. The decisions of the trader are driven by complex algorithms which are designed to follow a defined set of instructions in order to generate profits at a speed and frequency that is impossible for a human trader. The defined sets of rules are based on timing, price, quantity or any mathematical model.

It is estimated that 70% of trading in US stocks is done using. Lewis notes that:

By the summer of 2013, the world’s financial markets were designed to maximize the number of collisions between ordinary investors and high-frequency traders – at the expense of ordinary investors, and for the benefit of high-frequency traders, exchanges, Wall Street banks, and online brokerage firms.

Advocates of HFT will tell you that HFT provides liquidity and this means that the market has a lot of buyers and sellers which suggests that you can make trades without moving the price too much. A liquid market means that people will be more likely to invest. However there are those that worry about the liquidity of HFT as it could be illusory as it could disappear very quickly if stock prices collapse. Andrew Haldane of the Bank of England put it – the fear about this liquidity is that ‘in wartime, it disappears’. Furthermore, HFT has also produced huge swings in stock prices. On 6th May 2010 – know as the ‘Flash Crash of 2.45pm’ – the DJIA fell 9% in 5 minutes but then recovered most of that loss in the subsequent few minutes. But what is most worrying is that nobody can agree what happened because nobody had any control over it. It seems that we are writing things (algorithms) that we can no longer read. We should be worried about HFT as it reduces the amount of the quantity of real and valuable information in the stock market system. It make the system as a whole less stable and more risky. And it devotes an enormous amount of resources to an arms race that is of dubious value.

HFT and the real economy

A recent study of the commodity market found that up to 70% of all price movements in those markets didn’t correlate to events in the global economy. The price movements were driven by algorithms reacting to internal action in the market. This not only makes the market dumber but also a lot more unstable as humans find it impossible to oversee it – e.g Flash Crash of 2.45pm. If HFT traders add liquidity to the market then when the market crashed on 6th May they should have stepped in by buying falling prices of stocks. Turmoil in the markets is nothing new but the speed that it happens today makes trading harder to control raising systemic risk. Some companies will go to get great lengths to improve the speed of trades. In July 2010 a one-inch cable was completed to send a signal from Chicago to New Jersey at a cost of US$300 million. The improvements brought down the estimated roundtrip time of the signal from 13.1 milliseconds to 12.98 milliseconds. But when you are an algorithm 0.3 milliseconds is a long time. The billions of dollars that have been put into HFT over the last 6 years have only had a small impact on the ordinary investor. HFT looks like an arms race as it consumes an enormous amount of resource but generates very little social value and damages the market in the process.

Trading in the Fast Lane

Trading in fast laneAnother image from the NYT adaptation of Michael Lewis’ new book ‘Flash Boys’. In the microseconds it takes a high-frequency trader — depicted in blue — to reach the various stock exchanges housed in these New Jersey towns, the conventional trader’s order, theoretically, makes it only as far as the red line. The time differences — now under investigation by New York’s attorney general — can be financially advantageous in a number of ways. Credit Graphic: CLEVERºFRANKE. Data source: IEX.

It is the algorithms that outsmart humans not machines.

Algo TradingWhilst away on hockey tour in Malaysia I was able to avail myself of the ‘The Straits Times’ newspaper which is published in Singapore. One article that particularly caught my attention was that concerning the creativity of algorithms. Most are oblivious to their creativity yet highly sophisticated algorithms have created music based on the works of great artists but in a style that is personalised and therefore indicative of you the individual. They are also replacing writers – Professor Phil Parker of the Insead Business School in Paris has published more than a million reports on Amazon in just a couple of years. Using a proprietary algorithm that produces a report in 10 – 20 minutes instead of about 4 weeks. The algorithm pulls information from the web, performs econometric analyses, creates tables, formats the report and publishes it as a Word document. Professor Parker has also developed algorithms to produce poems, videos and video games.

Algo Trading
Although we could question the efficacy of algorithms on intangible dimensions such as “soul’ and “depth”, one area where they trounce human beings is stock trading. With up to 75% of trades on Wall Street done using computer programmes it is no wonder that algorithms execute trade at lightening speed and carry out numerous transactions every second. On the NYSE the average round-trip transaction time is 600 microseconds. To put into perspective if you blinked it takes you 300 milliseconds to complete the action – during that time NYSE executed 500 trades. This desire to improve efficiency in the market has led to extremely low costs of trading and very high stock liquidity. However it has also produced huge swings in stock prices. On 6th May 2010 – know as the ‘Flash Crash’ – the DJIA fell 9% in minutes but then recovered most of that loss in the subsequent few minutes.

The landscape of society was always made up by this uneasy relationship between nature and man. But now there is this third co-evolutionary force – Algorithms – and we will have to understand them as nature and in a way they are. Kevin Slavin Ted Talk