This told essay is based on a conversation with Barak Turovsky, former chief AI officer at Silicon Valley-based General Motors. He has also held executive positions at Google and Cisco. The following has been edited for length and clarity.
I have been working on AI and LLM since 2014. This was long before AI and LLM became the hottest thing on the planet.
I'm a former Google AI executive who led the first large-scale implementation of LLM and deep neural networks using Google Translate. I also worked as Chief Product and Technology Officer at a computer vision AI startup and as Vice President of AI at Cisco.
While I was at Cisco, I was approached by General Motors for the role of Chief AI Officer. It felt like a great crash course in developing physical products using AI. This role no longer exists as I left after GM restructured its software and AI organization. However, until November, I reported to the head of software engineering, and the head of software engineering reported to the CEO.
Some people may ask, “Is a dedicated AI person really necessary?”
You can call it something else, so ignore the title. However, I believe that successful AI implementation requires a leadership figure and commitment from the top to drive that change.
Functional business leaders such as CTOs and CIOs rarely or There is no understanding of AI. If you want to integrate AI at the software level, you need people with a different kind of expertise.
Traditionally large companies have powerful executives who want to reap the benefits of AI expansion, but are not necessarily in charge. Therefore, we need someone with deep knowledge of AI to guide traffic.
I would like to explain it by comparing it to a restaurant.
CAIO is like a master chef in a restaurant
This analogy is based on three main resources that create a product or dish. Initial resources include kitchen equipment or the AI infrastructure and models needed to build an AI solution.
The next thing to consider is the ingredients. This is data or internal assets used to train and run AI solutions. Lastly is the talent, or the restaurant staff. Different levels of expertise are required: busboy, short-order cook, sous chef, and master chef for a truly gourmet restaurant.
The complexity of creating the final product depends on the company's needs, especially whether the restaurant needs to prepare food in-house. For highly advanced, state-of-the-art models that can be considered the main course of a gourmet restaurant, companies often have to develop their own AI solutions, as standard versions may not be able to perform the required functions.
Think of CAIO as a master chef. They have to make sure that all the different parts work smoothly. If you're in an industry that requires state-of-the-art solutions, you also need to spend a lot of time making sure your most difficult deliverable: the main dish, comes out properly.
The hardest and most important part of the job is retaining the best people. Vendors will tell you that you can make a French soufflé in 15 minutes in a toaster oven. However, the materials always arrive late, the quality is questionable, and the quantities are inadequate. A customer comes in, declares he or she is hungry and wants to eat everything on the menu, and leaves midway through the meal without paying.
What CAIO should do
The details of each role vary. At GM, we've gone to the cutting edge of bringing AI to physical products like cars. It is largely untouched and attracts a lot of attention.
There are three buckets of what an AI chief executive should do. The first is AI human resources management. I focused on recruiting a top-class team. This is very important. Because the moment you step into a new field, you have just a fraction of your talent. They need to be motivated and flexible as they are still planning these areas.
Next, you need to build a culture of innovation throughout your company. You need to work with internal stakeholders who may be used to doing things a certain way, but need to change because of AI.
We also need to create organizational change. This starts with mapping your organization's needs and stakeholders. There are both AI enthusiasts and skeptics. In large organizations, they are not always easy to identify. You need to create a top-down and bottom-up framework with clear goals from the top.
In every role, you need to identify, develop, and empower champions. CAIO can't do all the magic while others just sit there.
