For the past few years, analysts, politicians, and other thinkers have fretted about the potential impact of a “robot revolution” which could see AI drive humans out of the workforce. The LA Times recently reported that robots could take over 38 percent of current human jobs within 15 years, and we have seen AI steadily move into new fields such as law or journalism which few had predicted before.
Despite the hype around AI, we will probably not see a world where humans are no longer needed. But as AI develops, businesses need to think about how to adapt to this changing paradigm. An examination of certain sectors should show both the benefits and limits of AI, and reveal that the future will be one where humans and AI cooperate instead of one supplanting the other.
Driving and mistakes
If there is any sector where AI could have the largest impact in putting humans out of work, it is driving. Last October, then-President Obama stated that the technology for driverless cars “is essentially here” and that there needed to be a “societal conversation” to protect low-wage low-skill workers. Millions of Americans routinely drive for their jobs, whether it is truckers driving as a career or ordinary people supplementing their income by delivering pizza or being an Uber driver, and seeing all those people go unemployed would be a major stress on society.
But while driverless cars are closer than ever, a crash last year by Tesla Autopilot shows the pitfalls of wholly depending on AI. The Tesla Autopilot crashed because it confused a white truck with a bright sky, a bizarre mistake which even a child could avoid.
That may seem like a silly, random mistake which is easy to fix. But a noticeable trend about AI is that they may commit fewer errors than humans, those mistakes are often incredibly wrong. The Watson AI which famously defeated humans in Jeopardy is a further example, as it would occasionally make incredibly obvious mistakes even though it won overall.
The overall trend of constant success punctuated by big mistakes should be a good example of both the potential and pitfalls of AI, especially when you consider the impact of a large mistake when driving a car. A business’s future AI plan should not necessarily view artificial intelligence as something which can replace humans entirely. Instead, humans need to play a supervisory role and be willing to jump in when an AI messes up.
The legal profession and machine learning
For years, an AI’s primary intelligence over a human brain is its ability to process huge amounts of information. Think of Deep Blue, which won over Garry Kasparov by being able to calculate so many more chess moves than a human mind. Many have thus believed that while AI would handle such brute force affairs, actual analysis and creativity would remain an exclusively human domain.
But there are routine, mundane parts of every business, which means that AI can play a vital role in any industry. The legal profession is one example, as the New York Times notes that law firms are using AI to scan and predict what documents will be relevant to a case. AI can cut away at much of the legal routine work and drudgery which junior lawyers used to do. Instead of having to memorize countless cases in order to do their jobs properly or conduct research, lawyers can be free to directly work with clients and negotiate deals.
The ideal, someday, is that an artificial intelligence could use machine learning to go through and memorize hundreds of court cases to “learn” to be a lawyer, and there has been major successes in machine learning in other fields such as medicine. But machine learning is not a cure-all which will make AI analyze things better than humans. In order for machine learning to work, an AI needs accurate data.
Unfortunately, too many businesses neglect data collection, which places them behind their competitors in responding to threats or opportunities. Any business planning out AI goals must understand that data is the lifeblood which AI depend on. This means that before figuring out AI development, a business should refine their data collection strategies and improve their actual human intelligence. Only then can a business’s new AI function at its peak capacity.
The potential of AI
Businesses everywhere are certainly salivating about what AI can accomplish, but we are a long way from the day when robots will supplant humans. Many aspects of many jobs can certainly be automated from skilled jobs like lawyers to unskilled jobs like lawyers, but humans will still have a part to play in this new economy.
Businesses should keep tabs on the latest in AI research in their fields and consider developing an AI plan. But basic things such as collecting data, emphasizing customer service, and just doing things well are still going to matter more both in the short and long term than AI.