Striking the AI talent balance

Posted on: March 14, 2019 by Philippe Poutonnet



Your AI project is underway.

  1. Desired outcomes: agreed upon and documented.
  2. Possible barriers: workarounds considered.
  3. Internal champions: engaged and leading the charge.
  4. AI team: what?

Don’t let this happen to you. Demand for AI workers has outpaced availability of those workers by 67%, according to new research from IDG sponsored by Google and Atos (access the research here). More importantly, only 24% of respondents have plans in place to fill that shortage.

Although this is a daunting statistic, I’m actually quite optimistic. The way I see it, there are only two hurdles to retaining and attracting talent to your AI program.

When you approach the topic of AI implementation in any IT organization, even the top performers inevitably wonder if they’re working to eliminate their own jobs. The first talent hurdle is to eliminate that uncertainty. Then you have to infuse new talent into your organization to keep it growing with new energy and knowledge. If you can do both – stabilize your team and build a good talent pipeline, you’ll strike the perfect balance of experience and fresh approaches.

Dispel the myths of AI in the workforce

Your team needs to understand that it’s not so much about job replacement but job transformation.

Think of all the AI tools already at your disposal. Things like spam filters and navigation systems make your life easier. Google recently released Smart Reply for mobile Gmail. It uses AI to generate quick-response options automatically, such as “thanks, got it” or “no, I don’t think so.” Already more than 10% of responses on mobile Gmail are using it. These tools gain popularity quickly because they make your life easier. As more business applications are similarly embedded with AI, humans gain time for critical and creative thinking.

AI tools and services are becoming easier and more accessible. There are a few hundred thousand data scientists in the world. They’re the privileged few who can use AI right now. Soon, however, we expect that every developer, data analyst and marketing analyst will be able to use drag-and-drop AI tools like Cloud AutoML, which is a codeless model training tool. Everyone in a company will be able to use it and build AI models to benefit their daily jobs.

Not only that, but AI will create new jobs. A decade ago, there were no social media managers or Uber drivers. Those and many other jobs didn’t exist. It’s not a degree in data science that will give you job security in the future. It’s adaptability and the capacity to reinvent yourself, which is difficult. If you can adapt to all the new things being thrown at you, you’ll really thrive in the future. Even at this stage in my career, I’m not done reinventing myself, though I am relieved that I might not have to do it many more times.

Today’s graduates need to be chameleons.

Win the AI talent war

It feels like all the AI talent goes to the Amazons, Microsofts and Googles of the world. To tip the scale in your favor, before you even start hunting, have in place your recruitment strategy, team structure, and job-switching opportunities.

If you’re looking for a software engineer on LinkedIn, it’s probably not going to work. Either they don’t have a profile set up or they’ve been picked up right out of school. Think of it like fishing. You’re not going to catch tuna at a nearby lake the same way you catch perch.

Senior AI researchers tend to come from academia. So you have to build relationships with nearby universities and schools, and be flexible in what you offer them. To fill more junior positions, you’ll have better luck with hackathons, recruitment events and grass roots initiatives. Think about where they are and go get them!

But you want to set them up for success from the beginning by establishing a team structure — short term, medium term and long term — before you hire. It takes a team of at least 6 people: data scientists, engineers and analysts. Many organizations will hire a few data scientist and place them in IT or software engineering. The new head of data science at one of these companies told me last month that before she was hired, the data scientists had been running wild. They were touching everything, trying to do too much and achieving nothing. Your AI team needs a good mix of doers and leaders with the vision of what you’re trying to achieve.

Another way to help your ML and AI talent succeed is to analyze the opportunities and motivations that lead them to change jobs. Then you can offer the incentives necessary to recruit people who would otherwise be out of reach. Companies must invest in helping them change roles from, say, data engineer to data scientist. Here at Google, we have a program called the Advanced Solutions Lab where we train some of our customers’ employees on AI. We have a dozen trainees right now. They are climbing a very steep hill to learn these skills. A few online courses won’t do. It definitely requires a commitment from the company to do this.

Hope is here

It is true that the majority of AI talent gets identified and recruited by the biggest companies during school. But there is hope! AI tools won’t be privileged much longer, and there are more candidates in schools and engineering programs.

Until the gap is filled, AI still needs a lot of hands-on customizations. It’s just too complex to manage on your own. Partner with specialists like Atos and Google who are willing and more than able to help. And keep an eye on AI advancements among your peers and competitors. You can start with the study we commissioned from IDG Research and CIO Magazine on the state of enterprise AI readiness. Access the full findings and more information at

Share this blog article

About Philippe Poutonnet
Director of Cloud AI Product Marketing at Google
Philippe Poutonnet leads the AI product marketing team and GTM for Google Cloud Platform. Philippe has spent his career helping companies build great products on the leading edge of what’s next, starting with his early days working at Forrester Research.

Follow or contact Philippe