War Games, Watson and the Flash Crash

In the 1983 sci-fi movie “War Games“, control over the launch of the US military’s nuclear missiles is given to a super-computer, with humans removed from the decision-making process for missile launch.

When the computer goes rogue and stops taking orders from its human operators, it is poised to start a global nuclear war by launching its missiles at Russia.  Armageddon is averted when the hero of the movie (a young Matthew Broderick) uses a simple children’s game (tic-tac-toe) to force the computer to assess the consequences of starting a nuclear war. The computer learns that there can be no winner and with this new information it makes a different assessment and halts the launch of the missiles.

You can watch the dramatic final scene of the movie here.

Blog 9 media wargames

Can modern computers make decisions, as the one in War Games appears to do?

Whilst some philosophers argue that computers can’t yet make decisions, most people would agree that computers can be programmed to do the following:

  • Gather and store large amounts of data, which is equivalent to human knowledge.
  • Apply algorithms or equations to that data in order to generate new information, which we can regard as being equivalent to forming judgements.
  • Use data, algorithms and models to evaluate different options and determine which option would be the best one, which is equivalent to making a decision.
  • Take actions on the basis of judgements or decisions it has made.

A real-life version of the super-computer in War Games is IBM’s Watson. Luckily for us, rather than nearly starting a nuclear war, Watson’s biggest achievement to date is winning the US television quiz show Jeopardy! in 2011, beating two of the show’s former winners.

However, IBM did not develop Watson just to win quiz shows. In February, IBM announced that Watson’s first commercial application would be providing recommendations regarding the provision of healthcare and healthcare services for lung cancer patients being treated at the Memorial Sloan–Kettering Cancer Center in New York. If this experiment proves successful, Watson could soon be present in many medical centres, providing recommendations for the treatment of many different diseases.

Handing decisions over to computers has already become common within the investment banking and investment management industries, which use computer programmes to automate the trading of shares and other investments. While this usually helps to make financial markets more efficient and more profitable for those companies that have access to such technology, it can sometimes cause problems.

In May 2010, the Dow Jones Industrial Average lost about 600 points in just a few minutes, a decline of 7%, equal to about $1 trillion in market value. Only a few minutes later this dramatic loss was reversed in an equally dramatic recovery back to the market’s original level. During what became known as the Flash Crash, the price of some major US companies’ shares fell to just 1 cent, whilst the price of others’ rose above $100,000. Subsequent investigations revealed that the automated trades of investment banks and hedge funds played a large role in causing this market anomaly. Among other things, this episode highlights that computers lack humans’ common sense.

Blog 9 media flash crash

Under what circumstances should we trust our decision-making to a computer, an equation or a model?

In her thought-provoking book “The Art of Choosing”, Sheena Iyengar, a professor at Columbia Business School and a well-known academic researcher on choice, tells the story of Howard Raiffa:

Once upon a time, former [Columbia University] faculty member Howard Raiffa, a pioneer in the field of decision analysis, was offered a position at Harvard, which was considered a step up in prestige. In an attempt to hold on to him, Columbia countered with an offer to triple his salary. Torn between the two options, he decided to ask his friend, a dean at Columbia, for advice. The dean, rather amused by the question, suggested that Raiffa use the techniques that had earned him the Havard offer in the first place: Break the decision down into its components, map the relationship between them, and do the math to determine which option was best for him. “You don’t understand,” Raiffa responded. “This is a serious decision.”

This story highlights that humans don’t generally make purely analytical, logical assessments of decisions, like a computer would perform a calculation. We typically consider the broader context of situations, the possible consequences for each of our options, and what seems sensible, and we are affected by emotions, expectations, assumptions, beliefs and biases.

For example, to return to investing in stocks and shares, a computer may be able to analyse market movements and make predictions for future movements; it may be able to tell me how to optimise my investment portfolio to achieve the best return for a given level of risk; and it may be able to make fast, efficient, profitable trades on my behalf. However, as an investor, it would still be necessary for me to make decisions about what level of risk was acceptable to me or what level of return I wanted to achieve. And my investment decisions could be influenced by:

  • Beliefs or assumptions; for example, I might think that solar energy is going to be the next big thing and so want to buy stocks in solar panel manufacturers.
  • Values; for example, I might not want to invest in any industries which I believe cause harm, such as tobacco companies or weapons manufacturers.
  • Qualitative judgements; for example, I might think that the CEO of Boxofrogs Inc is crazy and that he’s taking risks that will destroy the company (“Sell!”).

This article from the New York Times makes the case for “letting computers and people each do [the type of decision-making that] they do best.” It highlights that while computers are faster and more accurate for primarily data-driven, analytical decisions (and that they are immune to fatigue and don’t suddenly leave organisations, taking all their expertise with them), only humans are able to make decisions about the many “complicated human issues” within organisations and daily life.

Human life demands human decision-making.

Which subjects to study at school, which university to attend, which career path to follow, who to marry, how many children to have, what to call them, whether to move house, how many staff to lay off in a downturn, whether to put Grandma into a care home, when to retire, what to do in retirement, even whether to end our own lives should we become debilitated by disease or old age. The Nobel Prize-winning French philosopher Albert Camus said: “Life is the sum of all your choices.”

We are all faced with many complex human decisions and as the New York Times articles suggests, for human decisions, what’s important is to help people to think more clearly and to avoid making bad choices.

The ELECTIA approach to decision-making is designed to improve human decisions by taking human life and human nature into account. It respects and embraces the human nature of decision-making; that is, our aspirations, our ambition and our values, the practicalities of real life, and our human foibles and biases. Its approach is to encourage, inspire and guide our thinking and judgement through the use of concepts and questions. It does not provide answers, as an equation does; instead, answers come from the decision-maker or decision-makers and their deliberation.

Who really is the decision-maker?

The ELECTIA approach to decision-making does not preclude the use of other decision-making models, frameworks, or even computer simulations and programmes; it encourages the use of relevant supporting approaches to decision-making as part of the aspect Well grounded.

However, for human decisions, when we use models, frameworks or computers to support our decision-making, it is important that we remain clear that we are the decision-makers. It is important that we do not surrender our judgement and our responsibility for making good decisions to the decision-making approaches that we adopt. These tools can help, but only under very specific circumstances should we believe that any decision-making approach would provide an answer that can be taken and acted upon without any further consideration.

Returning to Watson and its use in diagnosing and treating lung cancer patients at the Memorial Sloan–Kettering Cancer Center, IBM says that Watson is intended only to provide decision-making support to doctors and nurses, not to replace them. Some observers disagree. However, even in a future with no human doctors or nurses, patients will still be human and that will require that some human decision-making is included in their care.

Maybe the patients themselves will be required to provide the human side of the decision-making for their medical treatment. If so, given their condition and difficult circumstances, it would seem to be more appropriate than ever that they receive some support to make good decisions on their own behalf, not just from powerful, analytical computers, but also from a decision-making approach which considers human nature.

I believe that the following quote from Sheena Iyengar’s “The Art of Choosing” summarises the human nature of decision-making very well: “[C]hoosing helps us create our lives. We make choices and are in turn made by them. Science can assist us in becoming more skillful choosers, but at its core, choice remains an art. To gain the most from it, we must embrace uncertainty and contradiction. It does not look the same to all eyes, nor can everyone agree on its purpose.”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s