Why You Don’t Need a Chief AI Officer, Now or Likely Ever. Here’s What to Do Instead

The Chief AI Officer–or CAIO–has emerged as one of the buzziest jobs in the business world as AI adoption accelerates. New CAIOs are typically tasked with the dual role of furthering business goals with AI while ensuring the tech has responsible governance. The CAIO, as envisioned, works with other C-suite leaders to evaluate new AI solutions, support product roadmaps, develop innovative AI offerings, implement responsible AI practices, and ensure all AI-impacted aspects of the business are running smoothly. However, for many organizations, a CAIO is often not the right way to become an AI-infused and AI-effective company.

The Journey to AI adoption & Issues That Remain

Let’s define a term or two and examine the hype. We’ve had model-based machine learning systems for decades, and they’ve been used effectively in a multitude of fields including medical diagnosis, fraud detection, and financial modeling. However, on November 30, 2022, the world of AI as the public knew it changed forever when ChatGPT became the first widely available public large language model (LLM). The world had shifted and would never go back.

In the past, machine learning and AI were seen as tools to be used much like compilers or text editors, with dedicated technology and personnel with special skills to execute them well–but Generative AI is different. LLMs–and the GenAI that they enable–offer capabilities that have never existed before. The technology is transformative in how it can help humans be better at things we already do well; after all, it’s trained on the corpus of human knowledge of things we’ve already done, and it’s really good at pattern matching and extrapolation. The good news is that there is a lot of work to be done on things we already do well: responding to customer support cases, synthesizing existing information into almost any form you want, summarizing calls and text, and much more.

(CKA/Shutterstock)

According to the AI hype cycle, for the first time in human history, we had a trusted partner that we could talk to, that could reason with us. Instead, we found out that, like your loud and opinionated uncle, AI could be just as convincing when giving you false facts as with the truth. Though bias, model drift and training quality had always been issues and still are today, that didn’t stop companies from integrating them across their organizations because the blast radius was controllable. With LLMs, that’s not the case. Humans love being told things, and when they are said convincingly, they believe them. So, while traditional AI model drift was damaging and had to be fixed, when a convincing AI is wrong about important things, we have a problem, Houston.

The Case for Shared AI Responsibility

With such broad impact and hype, it’s natural for C-suites and boards to wonder if they need somebody at the top whose sole job is to plot the path through the uncertainty. Enter the CAIO.

Many organizations jumped on the vision for a single person to steer their strategy, but there are more effective ways to accomplish their goals. What they should do instead: make sure that department-appropriate AI expertise is injected into almost every part of the company.

Imagine if when electricity came out (I wasn’t around then, despite what my kids think), every company had appointed a Chief Electrification Officer. Yes, every part of the company had the potential to be improved by electricity–but that doesn’t mean ownership should exist with one individual. The factory floor needed one plan, the service department needed another, and sales needed to understand how electricity was going to affect their own processes and customer needs. While flimsy strategies (“Electricity is good!”) can exist across an organization, actual execution requires per-department domain expertise, prioritization, and local leadership.

(FOTOGRIN/Shutterstock)

AI will affect companies on a smaller scale than electricity, but the impact will still be large and will influence operations across the entire company. Companies likely need a skilled practitioner giving advice on AI to the C-Suite regularly. But, if you put in a C-Suite executive, responsibility and accountability get mixed up.

There are many articles that take contrary positions to mine. They tout all the proposed responsibilities of the CAIO, but all I hear is the other C-Suite executives losing involvement, agency, and accountability for what is one of the most important initiatives they will lead in their career. They say fluffy things about how the CAIO is needed for competitive advantage and to have the company make better decisions, including “deploying AI” (whatever that means), to transform the business, improve customer service, etc. Last time I checked, these jobs already exist. In fact, while all of these things are needed, and AI is an integral part of improving all of them, having a central C-suite executive is likely harmful.

The Model for Effective AI Adoption and Integration

Don’t get me wrong; there is a big AI job to be done. In the CIO org, the use of AI in all the companies’ systems needs to be implemented well and with strong governance functions to ensure that models are used correctly and ethically. For example, we need to make sure that AI tools are effectively helping customers who ask for support, that Human Resources software responsibly uses AI, that Sales and Deal Desk have the right tools to summarize calls, analyze contracts, etc. and that the Talent Acquisition team is getting the benefits of AI while avoiding bias and promoting candidate diversity.

If the company produces technical products with a CTO, then it can make sense to have an AI platforms team, to make sure that AI is being used cost-effectively and consistently. The CMO of course needs to use AI products for analyzing SEO, creating documents, and analyzing competitive data. For software companies, GenAI can be a huge boost for both junior and senior developers due to its code generation capabilities. For the very few companies that are producing AI-tech products, they need to have an entire engineering and product team that are experts in AI.

Having one person oversee all of these functions is nearly impossible, and could (ironically) hamper AI operations and strategy, while slowing down business operations. Rather, it’s much more effective to empower C-suite leaders to embrace and utilize AI at their own discretion and pace, based on their department’s individual needs.

(pgraphis/Shutterstock)

However, if you do have a CAIO, or still feel like you want one, that’s ok. In that case, the CAIO should be in a role of advising and watching, different from any other C-Suite executive. Your other C-Suite executives are operators, not advisors. This person can be a source of advice for the board and the C-Suite on how effectively the company is adopting AI as well as identifying and deploying best practices throughout the company.

Whether you have a CAIO or not, an effective step toward successful AI integration and adoption is to implement an AI council. The council would monitor how AI is being adopted, and should include representatives from each department. Depending on a business and how it operates, the council would have representation from the organizations of the CIO, CTO, COO, etc. Each org would report out on their planned use of AI, what business benefits are promised, and how they will put cost and governance guardrails in place.

The CAIO (in a purely advisory, non-operational capacity) could be the chair of this council. Per the electrification example, this council would make sure that everybody was using electricity, leveraging it safely, and using the same plugs and voltages, for efficiency. The primary benefit of the AI council is to ensure that all voices are being heard, and that any AI decisions are a group effort, not made in a silo. It also lifts the burden from one person, who would be tasked with understanding all departments within an organization, and distributes that responsibility equally.

Takeaway

It’s an inescapable fact that AI, in both machine learning and GenAI, is transforming every company. AI is affecting your business, whether through external forces that reflect new needs and desires of your customers, competitors that are flanking you, or internal forces such as the need to raise efficiency, create better products, or have more predictability. You can choose to drive or be driven.

If you decide to drive, it’s important to do it in a way that respects the way your company and its departments currently function. All executives must have empowerment and accountability – giving them agency to make the changes and innovations they need and to customize the method and pace of AI adoption to their specific department. At the same time, just like in the rest of your company, you need coordination and governance, and you already have processes for those. Rather than creating new processes, incorporate AI adoption into those processes.

While AI is so new, you may need an AI council, or even a single highly placed advisor to help make the transformation. Over time, just like electricity, AI will become integrated into everything you do, and you won’t need a special position or special advice.

Good luck!

About the author: Mark Porter is the CTO at dbt Labs where he leads the engineering organization, including the development, research, and infrastructure teams, supporting mission-critical customers around the world, along with driving the future technical direction of the company. He has over three decades of experience at MongoDB, Grab, Amazon Web Services, NASA/JPL, Oracle, and other companies. In all of these roles, he has focused on building software for customers to use in their mission-critical businesses while also nurturing and growing excellent engineering cultures and teams. He began professionally programming at age 16 and is a named inventor on 15 patents. Porter has served on the boards of MongoDB and Splyt, and currently serves on the board of directors at GitLab. He holds a BS in Engineering and Applied Science from Caltech. When not at his keyboard, he spends time with his wife and 5 children.