The field of artificial intelligence has enjoyed tremendous progress over the past few years. Thanks to a number of advances, both technological and algorithmic, a new wave of AI applications is forming, turning what once belonged exclusively to futuristic sci-fi or dystopian stories into commoditized, ubiquitous technology.
Nevertheless, despite what you might have heard, AI is not quite ready to take over the world. While we are witnessing it achieve remarkable — and oftentimes superhuman — performance in tasks previously regarded as outside the bounds of computers, AI systems are still narrow in scope and remain, in essence, a prediction technology. Generally, machine intelligence is unable to adapt to similar but different tasks, barring substantial work and human intervention, and most crucially it lacks certain fundamental “general intelligence” traits such as deductive reasoning and the ability to learn how to learn.
Jobs, tasks, and automation
That is not to say that AI won’t transform our economy and job market. However, the process is gradual. Similar to the introduction of robotics on factory floors, in each step progressively more “human” tasks are delegated to capable autonomous systems, so little by little disruption occurs, new economies thrive, and old sectors fade away. At each stage of this inexorable automation march, tasks that were once believed to lie beyond the scope of computers and were reserved exclusively for humans and their “unique” abilities are nonchalantly redefined as not “real intelligence,” and thus the goalpost for “real AI” is moved a bit further. While this behavior can make us feel a little bit better, it certainly does not help with preparing for the rise of new technologies and the disruption that follows in their wake.
To be able to prepare for and take advantage of this machine intelligence-driven automation wave, it is important to first understand how automation takes place. If we break down a job as a combination of tasks and accept that automation is, according to Daron Acemoglu, the replacement of tasks previously performed by labor, it becomes clear that the process of automation happens in parallel to augmentation. Machines are first introduced to assist humans by augmenting their skills, which, in the context of AI and machine intelligence, is called intelligence augmentation (IA for short). Over time, as systems evolve, more tasks are shifted to autonomous agents, and eventually, machines can perform all but very few aspects of a job.
While the effects of this process are particularly important to those people in the job market, companies are also deeply affected by these changes. As a general rule, with new groundbreaking technologies comes new economic paradigms, and inevitably some companies thrive and others are left behind. As we continue to commoditize machine intelligence, predictions will get progressively cheaper, resulting in two fundamental changes. First, we will use predictions more broadly both for smaller tasks and for tasks where this was not an option before. Second, the value of goods that complement predictions will thus rise while the value of those that substitute it (such as labor) will fall.
As your business fights to adapt to this new paradigm, it is imperative that you identify which tasks — whether part of your internal operations or related to the products you offer — you will automate. In other words, how can machine intelligence augment your employees and/or customers? A good rule of thumb to answer this question is: In situations where the only information required to make the decision is what is immediately available (i.e., the signal itself), machine intelligence can potentially take over.
Consider the following scenarios:
- A customer representative handling a support ticket regarding a common problem usually solved by taking the same predetermined set of steps.
- A salesperson going through a long list of leads trying to determine which are qualified leads based on a number of features.
- A doctor analyzing a single x-ray in search of visual cues to support a diagnosis.
All three are examples of situations for which the information immediately available is sufficient to make a decision. In these scenarios, a business could augment all three professionals — hence becoming more efficient — with AI-powered technology.
It is important to emphasize that the augmented worker remains indispensable. AI is still unable to handle complex support cases involving multiple pieces of information. No AI will pick up the phone and have the ability to talk to a potential customer and close a deal. And a doctor is much more than an x-ray analysis machine. Notwithstanding, in all the above scenarios the professional could enjoy an efficiency boost by delegating a portion of their tasks to a machine.
We can apply this rule of thumb to any job or any company to assist in identifying new opportunities for the use of AI. If you imagine for a second that predictions are cheap and accurate, and you can automate repetitive and narrow tasks, how can that affect your job or the way your company operates? By asking this question and using the rule of thumb of automation as a guide, you may be surprised by the number of potential tasks around you that are prime candidates for intelligence augmentation. Here are some examples:
Lead scoring: Instead of having sales professionals spend time determining which leads they should prioritize by analyzing the same set of parameters over and over, they could use machine intelligence to quickly generate a short list of qualified leads.
Customer service: Typically, a customer service department deals with a wide spectrum of cases, from rare and complicated to common and simple. A business could delegate the latter to an AI system, freeing up agents’ time and allowing them to focus on harder and more time-consuming cases.
Optimizing marketing strategies: Data is ubiquitous in the marketing world. Unsurprisingly, the role machine intelligence plays in helping with decision making has grown steadily. From optimizing channel mix to helping with experimentation, AI allows marketers to spend more time thinking about their customers and less time tinkering with campaign optimization.
Infrastructure load: The ability to anticipate server load and take appropriate action can help prevent downtime, save costs, and decrease the time system administrators spend dealing with unforeseen server problems.
Sketch to product: Going from idea to prototype with the help of AI almost immediately empowers designers and products managers to test their hypothesis faster, leading to a highly efficient product-development cycle.
This list contains a few of the myriad possible applications a company can find for machine intelligence.
The era of AI-powered automation is here to stay. Preparing, adapting, and, most importantly, taking advantage of it are key steps for any business or individual interested in staying ahead of the curve. In order to do so, it is important to understand what aspects of your business are more prone to intelligent augmentation and take measures accordingly. One thing is certain: Companies that fail to embrace new technology paradigms are left behind, wondering how their competitors can be so efficient and move so quickly.
Mazdak Rezvani is the founder and CEO of Chatkit, an artificial intelligence conversational marketing platform that enables brands to manage and automate conversations with their customers on messaging channels such as Facebook Messenger.