Solving The Skills Gap With User-Friendly Artificial Intelligence

• Updated on: 2019-04-05
• Read original article here

While taxes dominated the headlines coming out of Davos World Economic Forum this year, I believe the tech world’s real focus remained on artificial intelligence (AI), automation and the associated skills gap. In an interview with CNBC, Workday CEO Aneel Bhusri said, “AI technology is in play and companies are going to use it, or else they won't remain competitive. But it's going to create a skills gap.”

Of course, we’ve been talking about skills gaps for years. And while I’m inclined to agree with Bhusri about retraining workers whose skills are automated by AI, I think the shortage of AI experts is a much more difficult problem to solve.

In 2018, Indeed Hiring Lab reported that demand for AI skills more than doubled over the preceding three years. In that same report, they demonstrated that interest in these jobs had leveled off. I've observed that many businesses have entered the war for AI talent. So what’s their game plan? It appears they are going to pay enormous salaries to AI experts.

If you’re a large company, that plan might work. But how long can smaller businesses sustain that? How long will it be before this results in either unsustainable business models or noncompetitive markets?

Both through my own company and through mentoring others, I have experienced what it means to build AI that has the human who will be using it in mind while it’s being built. I believe the answer to the skills gap isn’t to throw more humans at the problem, but instead to throw better AI at the humans.

How to vet and create user-friendly AI.

I've observed that much of the AI being built, bought and sold today lacks a user interface. The technology industry tends to swing back and forth between front-end and back-end innovation. Today, we’re in the late stages of a back-end innovation cycle. Silicon Valley has been focused on the infrastructure, databases and all the middleware it takes to stitch together increasingly complex systems. From my perspective, no one seems to be worried about how we empower people to better leverage AI on a daily basis.

1. Get all the right people in the room.

Don't wait until you have a minimum viable product to ask for your team’s feedback. You need designers, strategists, data scientists and technologists in the room so everyone is aligned on the business needs, implementation challenges and usability concerns from the get-go. AI should be a tool that amplifies the work of your entire team, not solely the experts who are building it. Consider this: You’ve hired a marketing team presumably because they are equipped to do the job. While the allure of marketing technology is strong, the goal of the technology isn’t to replace them; it’s to help them do their job better. Unless you are in their day-to-day, the only people who have insight into how to do marketing better are the marketers themselves.

2. Define how your product strategy leads to execution.

With so many voices in the room, it’s far too easy to end up with an inelegant product that satisfies ideas without solving your business’s most critical needs. Identify these imperatives, and translate insights about processes into actionable, deliverable solutions. Building AI is no simple task, so prioritize how products and services will be launched, and make sure the infrastructure exists to create a continuous feedback loop between all stakeholders. This means making sure you have the right stakeholders in regular communication through meetings or other means and knowing what questions need to be asked at every step to make sure you’re still moving in the right direction. Don’t make it a side project – carve out the time you need each week to make sure you’re giving it the attention it deserves.

With the amount of time, energy and resources you’re pouring into the development of an AI solution, you need to do everything possible to make it easy to refine and perfect over time. If the last decade of innovation has taught us anything, it’s that the future is closer than we think, and we need to be prepared to facilitate the integration of new technologies. When choosing a vendor, ask the hard questions: “What if I want to change this in six months, one year or five years? What are my limitations?” If they don’t have a good answer for exactly what you can and cannot do, move on.

The great irony of automation is that machines are best suited to take the math-intensive and highly specialized jobs of their creators. Yes, I believe you should invest in AI to remain competitive, but I don’t believe growing the AI talent pool is the only way to do it. Machines have the skills to close the skills gap, and humans have good reason to expect more from their machines. But in my opinion, AI systems need to pay more attention to end users’ needs.

While taxes dominated the headlines coming out of Davos World Economic Forum this year, I believe the tech world’s real focus remained on artificial intelligence (AI), automation and the associated skills gap. In an interview with CNBC, Workday CEO Aneel Bhusri said, “AI technology is in play and companies are going to use it, or else they won't remain competitive. But it's going to create a skills gap.”

Of course, we’ve been talking about skills gaps for years. And while I’m inclined to agree with Bhusri about retraining workers whose skills are automated by AI, I think the shortage of AI experts is a much more difficult problem to solve.

In 2018, Indeed Hiring Lab reported that demand for AI skills more than doubled over the preceding three years. In that same report, they demonstrated that interest in these jobs had leveled off. I've observed that many businesses have entered the war for AI talent. So what’s their game plan? It appears they are going to pay enormous salaries to AI experts.

If you’re a large company, that plan might work. But how long can smaller businesses sustain that? How long will it be before this results in either unsustainable business models or noncompetitive markets?

Both through my own company and through mentoring others, I have experienced what it means to build AI that has the human who will be using it in mind while it’s being built. I believe the answer to the skills gap isn’t to throw more humans at the problem, but instead to throw better AI at the humans.

How to vet and create user-friendly AI.

I've observed that much of the AI being built, bought and sold today lacks a user interface. The technology industry tends to swing back and forth between front-end and back-end innovation. Today, we’re in the late stages of a back-end innovation cycle. Silicon Valley has been focused on the infrastructure, databases and all the middleware it takes to stitch together increasingly complex systems. From my perspective, no one seems to be worried about how we empower people to better leverage AI on a daily basis.

1. Get all the right people in the room.

Don't wait until you have a minimum viable product to ask for your team’s feedback. You need designers, strategists, data scientists and technologists in the room so everyone is aligned on the business needs, implementation challenges and usability concerns from the get-go. AI should be a tool that amplifies the work of your entire team, not solely the experts who are building it. Consider this: You’ve hired a marketing team presumably because they are equipped to do the job. While the allure of marketing technology is strong, the goal of the technology isn’t to replace them; it’s to help them do their job better. Unless you are in their day-to-day, the only people who have insight into how to do marketing better are the marketers themselves.

2. Define how your product strategy leads to execution.

With so many voices in the room, it’s far too easy to end up with an inelegant product that satisfies ideas without solving your business’s most critical needs. Identify these imperatives, and translate insights about processes into actionable, deliverable solutions. Building AI is no simple task, so prioritize how products and services will be launched, and make sure the infrastructure exists to create a continuous feedback loop between all stakeholders. This means making sure you have the right stakeholders in regular communication through meetings or other means and knowing what questions need to be asked at every step to make sure you’re still moving in the right direction. Don’t make it a side project – carve out the time you need each week to make sure you’re giving it the attention it deserves.

With the amount of time, energy and resources you’re pouring into the development of an AI solution, you need to do everything possible to make it easy to refine and perfect over time. If the last decade of innovation has taught us anything, it’s that the future is closer than we think, and we need to be prepared to facilitate the integration of new technologies. When choosing a vendor, ask the hard questions: “What if I want to change this in six months, one year or five years? What are my limitations?” If they don’t have a good answer for exactly what you can and cannot do, move on.

The great irony of automation is that machines are best suited to take the math-intensive and highly specialized jobs of their creators. Yes, I believe you should invest in AI to remain competitive, but I don’t believe growing the AI talent pool is the only way to do it. Machines have the skills to close the skills gap, and humans have good reason to expect more from their machines. But in my opinion, AI systems need to pay more attention to end users’ needs.