How Artificial Intelligence Can Break a Business in Two Minutes

May 4, 2017
Vote on Hacker News

BB&T, Wells Fargo & Co and several other leading financial giants are not the nimble, agile businesses that many other industries are home to. They are slow, deliberate, and methodical. Every risk is weighed, every reward scrutinised. That’s why it made headlines when it was revealed that they were, from this quarter, diverting considerable resources into developing and using artificial intelligence for data mining within their organisation. They’re even looking into Blockchain as a contract management and financial transaction solution. Even slow movers like banks are now getting on the AI and general tech bandwagon. If there was ever a sign of AI’s rising success, this is it.

There are dozens of ways in which the rise of Artificial Intelligence is making possible things in business which no human could realistically achieve – at least not while maintaining profitability. From route planning to business intelligence to the automation of manual processes and more, artificial intelligence has allowed many businesses to grow far beyond what they could have done without it. One of the biggest examples is Data Mining. Data is known in the IT industry as ‘new oil’ because it holds immense monetary value both in potential earnings and actual physical value for the data itself. The ability of artificial intelligence to pore through massive quantities of data to reach meaningful conclusions allow businesses to act in response to business intelligence they would not have had previously.

As an example of making fast money with artificial intelligence, it is currently being used by stock traders to make snap trading decisions in response to news reports. Huge natural disaster in an area which exports large amounts of timber? Bam. Timber stocks are bought by the AI, and in minutes the price will go up in response to a perceived shortage as human sellers react. These stocks can then be sold for a quick profit. This usage of neural networks and genetic algorithms to manage stock portfolios isn’t anything new, but the complexity and autonomy of neural networks is increasing on an almost day-by-day basis. It almost makes a mockery of Moore’s law.

So how can this miracle tool possibly cause damage to a business? That last example probably rang one or two alarm bells in your head, and for good reason. Like humans, artificial intelligences can make mistakes when they received incorrect of misleading data. One of the most successful terror attacks of all time went largely unnoticed by the general public because it harnessed this exact fault to cause huge financial damage to the United States trade markets.

An Associated Press twitter account was hacked on April 23, 2013, and tweeted the following message:

‘Breaking: Two Explosions in the White House and Barack Obama is Injured’

This tweet, a mere eight days after the infamous Boston Marathon Bombing, was a fake. There had been no attack, and within minutes, as frantic phone calls were made around the world, it was confirmed that President Obama was safe and nothing was amiss. The news was publicly dismissed the same day.

So what does this have to do with artificial intelligence? Within a mere second of the tweet going live, dozens of high-frequency trading artificial intelligences in trading markets throughout the USA read the tweet automatically and extracted data. They understood that there had been an attack and anticipated a sudden, drastic change in the stock markets. They tried to get ahead of the game and all acted in unison in response to this made-up assassination attempt. Within the two minutes it took a human to ‘wrestle the controls’ back from the AI, $136 billion worth of damage had been done to the US stock market.

The culprits of this tweet were the Syrian Electronic Army, a cyberterrorist hacker group supported by the Syrian government. A human being would have seen this tweet and verified the data before risking quick trading decisions. The artificial intelligence did not. This shows one of the major weaknesses in handing over the controls to AI – you stand to gain a lot of money, but you also stand to lose a lot more.

Now consider that this AI was not in charge of the stock market. What if it was in charge of the controls in a plane? Or a production line at a factory? Reacting to false information or misinterpreting information from around it could be catastrophic, especially if it doesn’t realise its mistake.

On the less damaging but no less worrying front, Microsoft released their most intelligent chat-bot AI yet named ‘Tay’ onto the internet in March 2016. Within 24 hours, this innocent ‘teen-girl-themed’ robot proclaimed a love and admiration for Adolf Hitler, proclaimed that George Bush was responsible for the 9/11 terror attacks and a plethora of other utterly insane statements. Unfortunately, lots of intelligent AI is intelligent precisely due to machine learning. It learns from those it comes into contact with. Anyone who has spent more than five minutes on the Internet knows full well the kind of people that dwell there. This is another weakness of AI in business – it’ll work great if the people and things that it interacts with are whole, family-friendly influences. Less so if they’re not. An AI that interacts with the public a lot is a double-edged sword.

Machine learning’s unfortunate downsides can bring ruin upon an unsuspecting business. You can’t vet the data because of the large data requirements. Accidental unsavoury data aside, the algorithms might not even work properly because it’s very hard to optimise machine learning for the ‘real world’ of human beings, and all of the randomness that contains. In fields like FinTech, it can be relatively easy to evaluate how a self-teaching AI is ‘doing’ in regards to your goals. However, in less ‘concrete’ sciences like human interaction or biological medicine (as opposed to chemical) there is no ‘right’ or ‘wrong’ so it is hard to evaluate on an ongoing basis.

Though more businesses jump on the A.I. and modern-tech bandwagon every day and success stories flood the headlines, it is important to remember that it’s not always fat profit margins and shareholder high-fives. Introducing technology like AI into business can sink your business as soon as take it to new horizons. Technology is like fire: use it wisely or get burned.