Artificial intelligence (AI) has been around longer than most people realize. The intent behind much of AI is to free us from mundane repetitive tasks, giving us more time to grow our intellects and businesses, with more interesting, evolving actions. We want what we want when we want it. AI offers us that access with speed and accuracy when we need it.
In London, self-driving robots deliver food. In Pasadena, California, a robot named Flippy can cook it. Last fall, an autonomous train made its way across the Australian outback for the first time, and Zhuzhou, China, began testing a trackless and driverless train that navigates city streets by means of lines painted on the road. From writing articles for The Washington Post to creating music, artificial intelligence is everywhere. And its adoption is rapidly becoming necessary for businesses to stay competitive.
How does this affect human employees? As co-founder of a company that utilizes artificial intelligence to provide customer support solutions, I believe that low-skilled jobs are most likely to be affected and most chances of being automated. White collar jobs are also at risk though with AI taking a bigger role in the financial industry.
But despite all this, the future for human employees may be much brighter than many recent predictions. While AI destroys jobs, it also creates them. And according to a report from the research firm Gartner, artificial intelligence is currently creating more jobs than it destroys, with a net increase of over two million jobs by 2025. This includes not only the obvious jobs such as software engineers but also low-level jobs such as training AI to recognize objects or human activity and many others.
AI may destroy jobs and it may create them, but it's not always about man versus machine. AI can be at its best when it helps humans to perform jobs. For example, last year, Walmart announced it was beginning tests of shelf scanning robots at 50 locations. These robots are not intended to replace human workers but to make them more efficient. The robots free employees from the tedious task of walking the aisles looking for out of stock products and allow them to focus their time on filling the shelves, replacing items left in the wrong place and fixing problems that the robots notify them of. The goal here is to reduce the number of times a customer looks for an item only to discover an empty shelf.
In the pharmaceutical industry, artificial intelligence can take on tasks that human minds simply can't do. According to a study from Tufts University, it can take over a decade and cost over $2.5 billion to develop a drug from start to approval and market. However, most drugs don't make it to market, some failing early, but others failing close to the end when years and millions or billions of dollars have already been spent. AI can leverage the vast amounts of data regarding medicine and health, thus potentially lowering the rate of failed trials. It can also help find appropriate patients to participate in clinical trials, model the behavior of molecules to help predict how they will behave in the human body, and find genetic biomarkers that allow medicine to be tailored to individuals.
Artificial Intelligence Needs Humans
In the above examples, artificial intelligence plays a part in preventing human errors. However, AI also still needs human oversight to prevent its own errors.
In July of last year, a bot designed to create phone cases based on popular image searches went terribly wrong and began creating cases with disturbing medical imagery and inappropriate images which were listed for sale on Amazon by the third party seller. In a far less amusing example of AI gone wrong, it took less than a day for Twitter to teach Microsoft's AI account "Tay Tweets" to spout racism, sexism and love for Hitler. To prevent such malformed sentences, AI models need more training data and a proactive human oversight.
The stakes grow far more serious when AI operates heavy machinery or is involved in healthcare. Autonomous vehicles have been lauded for their potential to reduce collisions, 94 percent of which are caused by human error, according to the NHTSA. After all, autonomous vehicles won't drive while drunk, tired or distracted. However, autonomous vehicles have already failed to prevent two deaths despite the presence of safety drivers. The safety driver of a 2016 Tesla S may have relied too heavily on the autopilot. Data showed that he ignored seven warnings to return his hands to the wheel before the vehicle failed confused the white of a tractor-trailer for open sky and drove right into it. Though we all believe, the autonomous vehicles would become the norm in the next decade, safety regulations and substantial human oversight are very much needed and will be needed for the foreseeable future.
From exploring places humans can't go to finding meaning from sources of data too large for humans to analyze, to helping doctors make diagnoses to helping prevent accidents, the potential for artificial intelligence to benefit humans appears limitless. There is valid concern that even as AI saves lives and helps businesses thrive, it will destroy livelihoods. Without a doubt, AI is taking over jobs once done by humans. However, it also creates jobs, and AI needs people to train it and watch over it. At its best, AI works with people instead of in place of them -- removing the tedious parts of jobs so employees can focus on better things, doing tasks that humans were unable to, and helping employees better do their jobs.