“Those who master the game of algorithms are winning the game of business by an absolute landslide.” Salim Ismail
Q: Algorithms are:
- A step-by-step method for solving problems or completing tasks
- The key to extraordinary business success and profits
- A potential threat to our careers, economy, and democracy
- Reflections of our own biases
- Increasingly making decisions formerly made by humans
A: All of the above, and more.
To many of us, the term algorithm calls to mind a future where many human decisions will be ceded to sophisticated machines. In some cases, algorithms already determine whether we get a loan, how much interest we pay, which online ads we see, and whether or not we are hired. Indeed, algorithms are now taking on more legal research tasks and are being used to predict future crimes. And as machine-learning algorithms gain popularity, it seems that their use will multiply even further as current algorithms gain the capability to program themselves and write new algorithms.
The profound growth of algorithms
Algorithms likely were first used in Egypt in approximately 2,000 BC as a system to help standardize the way ancient scribes multiplied numbers. The profound potential of algorithms was recognized in the 1940s by computer science visionary Alan Turing. Since that time, algorithms have grown in power and influence with the rise of affordable computing and the oceans of data we humans generate each day with our computers and smart devices.
Although algorithms work primarily behind the scenes, their impact—both positive and negative—is growing along with their power and influence. Algorithms are incredibly useful in helping us make sense of the massive datasets we work with each day. In fact, algorithms may help us unlock the answers to many of the world’s biggest problems, which we call global grand challenges. You might not realize it, but you’re already using algorithms every time you:
- Send an email
- Shop on Amazon
- Perform a Google search
- Navigate with Google Maps
- Read your Facebook feed
- Browse movies and TV shows on Netflix
- Fly in a commercial aircraft
- Drive through an urban area
- Trade securities online
- Interact with smart speakers
Today’s commercial aircraft increasingly are guided by algorithms, while dynamic traffic control system algorithms help to relieve urban congestion. If you do a Google search to learn more about algorithms, that search would be powered by PageRank, an algorithm used to rank web pages and return relevant search results.
There’s no doubt that algorithms can make our lives simpler and more convenient. But there’s also increasing concern that the potential downsides of an algorithm-driven world may be overlooked as we rush to embrace their commercial benefits and convenience.
The algorithmic future: massive potential and massive risk
“Fact: We have already turned our world over to machine learning and algorithms. The question now is, how to better understand and manage what we have done?”—Barry Chudakov
Algorithms are used extensively in financial markets. Around 70% of overall U.S. trading volume is automated and managed by algorithms following the millions of daily movements in equity markets and responding millions of times more quickly than human traders.
What could go wrong? A good example is the ominously named “flash crash.” Perhaps the most famous incident occurred on May 6, 2010, when the Dow Jones Industrial Average suddenly plummeted more than 1,000 points and then rose nearly as quickly to its previous value within minutes. Although it was unknown at the time, it’s generally agreed that the 2010 crash was caused by a single individual using spoofing algorithms, which are automated programs used to bid up the price of a security before canceling the transaction. The trader used the spoofing algorithm to create the appearance of a massive market sell-off that in turn prompted others to sell, enabling him to buy those securities at a significantly reduced price.
On April 21, 2015, the US Department of Justice charged Navinder Singh Sarao, an individual trader working from his home outside London, with 22 criminal counts including fraud and market manipulation through the use of spoofing algorithms. Sarao eventually went to jail on charges carrying a maximum sentence of 380 years.
Another famous example of algorithms gone awry is Microsoft’s Twitter bot named Tay that was presented as an experiment in “conversational understanding.” Unfortunately, the conversations quickly took a wrong turn as Tay, prompted by Twitter trolls, began to respond with a variety of misogynistic, racist, and fascist messages. Microsoft quickly shut down Tay and deleted several of its more offensive tweets. Although the experiment spawned a wave of caustic humor from the Twitter community, it also raised real concerns about an algorithm that developed offensive biases so quickly, with no such intent from its programmers.
Algorithms for good, or evil?
“The problem is that algorithms know so much about us while we know so little about them.”—Jacob Brogan
Despite the growing impact of algorithms on our lives, it’s too early to predict whether the net effect will be positive or negative. Research by the Pew Internet Trust showed that almost all respondents acknowledged some potential negatives concerning algorithm-based decision-making, but beyond that, the results pointed to a split verdict. According to the research:
- 38% of respondents predicted that the positive impacts of algorithms will outweigh the negatives
- 37% said negatives will outweigh positives
- 25% said the overall impact of algorithms will be about 50% positive and 50% negative
What is certain is that the algorithmic economy continues to grow at a remarkable pace. The emergence of self-learning algorithms that are used to write others means even more powerful and sophisticated algorithms are on the way.
It’s another example of how we’re reaching the future faster, driven by emerging and converging technologies. While the net effect of algorithms remains to be seen, one thing is certain: accelerated technology requires accelerated learning.
The more we learn about the nature and applications of algorithms, the more we can tip the scales toward the positive. Education and understanding, along with intelligent governance, are our best bets to realize the immense potential benefits of algorithms and avoid creating technology we can’t understand or control.