History offers many examples of technologies that remade the workplace by eliminating jobs and mechanizing entire industries. Still, few “disruptors” have bred angst quite like artificial intelligence (AI).
AI refers to the ability for computers to act independent of human instruction. Informed through machine learning, AI couples the ability to recognize patterns in data and thus make accurate predictions. It also involves the ability to recalibrate, learning from missteps. The implications are vast. But unlike past technologies, AI is really just beginning to make its way into the workplace, and already many are in a frenzy over the impact.
Select business advisories have taken to declarations like AI solutions are “fundamentally reinventing how businesses run, compete and thrive.” Meanwhile, critics of AI have taken to conjuring darker visions of robot overlords.
One of the most well known examples of AI is Watson, the IBM computer system that leapt to fame by winning the game show Jeopardy in 2011. Since then, there have been a series of TV commercials with Watson cast as a sort of philosopher-king of technology paired alongside the mortal likes of Bob Dylan and Serena Williams. However, getting Watson to solve real world problems has produced mixed results at best. This year M.D. Anderson Cancer Center ended its partnership with Watson Health.
It seems Watson was overbilled. The MIT Technology Review noted that the present chorus of doubters regarding Watson are not so much chiming about the failure of AI technology, but rather expressing “a reaction to IBM’s overly optimistic claims of how far Watson would be by now.”
More generally, AI does not appear to be the technology that will reshape business in the near term. According to a 2017 survey conducted by KPMG (.pdf), when it comes to technologies that will drive business innovation over the next three years AI comes in third, behind the Internet of Things (IoT) and robotics. However, as the report notes, it is likely to be the convergence of these technologies together that has the most disruptive impact going forward.
Like other disruptors, the slow immersion of AI into the world owes to economic reality. Today the cost of deploying AI solutions remains relatively high, and it can be difficult to find people with the right skills. Harnessing AI’s predictive power depends on costs coming down and making it easier for companies to implement into real-world situations. As the Wall Street Journal’s Greg Ip noted in a recent column, “Treating prediction as an input into an economic process makes it much easier to map AI’s impact.”
However, despite expressing caution about the dramatic promises some analysts are making about AI, for the software development industry there will be some major changes coming our way. It promises to impact both software development and testing. Thanks to its ability to rapidly sift data and assign probabilities to outcomes, AI will significantly improve software testing and feedback, as well as help developers create better software using AI-based functionality.
Moreover, the path to the AI era is not a cobblestone of mystical skills. Machine learning, and thus AI, both rely heavily on programming with Python, because of its ease of use with data libraries. The biggest change is rather the need for software developers to change their mindset when working with AI.
Rather than fearing AI, software firms in particular are well situated to benefit from the gains in testing efficiency, and improvements in software development, that it promises.