I'm the former chief AI officer at GM. Being the CAIO is like being the master chef of a restaurant.

• As General Motors' first chief AI officer, Barak Turovsky hired talent and mapped organizational change.
• The CAIO role is growing in popularity as companies look to effectively implement AI tools.
• Turovsky compared being CAIO to being a master chef, who needs to make sure all parts of a restaurant run smoothly.

This as-told-to essay is based on a conversation with Barak Turovsky, the former Chief AI Officer at General Motors, based in Silicon Valley. He also held executive roles at Google and Cisco. The following has been edited for length and clarity.

I have worked on AI and LLMs since 2014 — way before they became the hottest thing on Earth.

I'm an ex-Google AI exec who led the first scaled deployment of LLMs and Deep Neural Networks with Google Translate. I also worked as the Chief Product and Technology Officer at a computer vision AI startup, and as the VP of AI at Cisco.

General Motors approached me for the Chief AI Officer role while I was at Cisco, and it felt like a great crash course on using AI to develop physical products. The role no longer exists because I left after GM restructured its software and AI organization; however, until November, I reported to the head of software engineering, who reported to the CEO.

Some people ask, "Do you really need a dedicated AI officer?"

Let's ignore the title because you can call it different names, but I do believe successful AI implementation requires someone in leadership to drive that change, as well as commitment from the top.

Functional business leaders, such as the CTO or CIO, may have little or no understanding of AI. If you want to integrate AI on a software level, you need someone with a different kind of expertise.

Traditional large companies have powerful executives who want to own the benefits of scaling AI, but not necessarily the responsibility. Therefore, someone with deep AI knowledge is needed to direct the traffic.

I like to use a restaurant analogy to break it down.

A CAIO is like a master chef at a restaurant

The analogy is based on three primary resources that create products, or dishes. The first resource includes the kitchen equipment, or the AI infrastructure and models necessary to build AI solutions.

The next can be thought of as the ingredients. It's the data or internal assets used to train and run AI solutions. The last one is talent, or the restaurant staff. You need expertise at different levels — busboys, short-order cooks, sous chefs, and master chefs for the really gourmet restaurants.

The complexity of creating the final product depends on the company's needs, specifically whether the restaurant needs to prepare the food internally. For very advanced, cutting-edge models, which can be thought of as the main course at a gourmet restaurant, companies often need to develop their own AI solutions because standard versions may not perform the required functions.

Think of the CAIO as the master chef. They need to make sure all the different pieces run smoothly. If you are in an industry that requires cutting-edge solutions, you also need to spend a lot of time making sure that the hardest output — a.k.a. the main dish — comes out just right.

The hardest and most important part of the job is securing top talent. Vendors will tell you that their toaster ovens can pop out a French soufflé in 15 minutes. Yet your ingredients will consistently arrive late, of dubious quality, and in incorrect amounts. Your customers will come in, declare they are hungry and want to eat the whole menu, and then leave mid-meal without paying.

What CAIOs should do

The specifics of each role vary. At GM, I worked on a cutting-edge area because AI for physical products, like cars, is largely untouched and getting a lot of traction.

There are three buckets of what a chief AI officer should do. First is AI talent management. I focused a lot on hiring a top-tier team, which is very important because the moment you enter a novel space, you have a small sliver of talent. They need to be motivated and flexible because you're still mapping out those areas.

Then, you need to create a culture of innovation for the company in general. You need to work with internal stakeholders who might be used to doing things in a certain way, but need to change because of AI.

You also need to create organizational change, which starts with mapping the needs and players of your organization. You have people who are AI enthusiasts and skeptics. In a large organization, it's not always easy to identify them. You need to create a top-down and bottom-up framework, which includes clear goals from the top.

In every function, you need to identify champions, and you need to nurture and empower them. The CAIO can't do all the magic while everyone else just sits there.