These days, it seems like everyone is becoming an “AI expert” by doing some prompt engineering or experimenting with AI tools to help create music. Everyone now considers themselves an AI developer. But not too long ago, the worlds of AI users and AI developers and engineers were very different.
How do you structure your AI team?
As organizations increasingly rely on AI, who are the real members of the AI team? A recent AI Today podcast explored this topic of evolving AI teams. Before LLMs became all the rage, the worlds of AI users and AI developers and engineers were very different. Just a few years ago, if you wanted to do something with AI, you needed highly skilled, well-paid, and hard-to-find people. These are the so-called “unicorns'' that companies have been searching for. Companies were unable to hire data scientists quickly, and the search for this talent was intense.
Companies needed these highly skilled data scientists, machine learning engineers, data engineers, and other high-paying, hard-to-find talent. These are hard to find because you can't become a data scientist or machine learning engineer by just going to a code academy for a few weeks or reading something online. It took me a lot of time to master those skills.
But now, thanks to generative AI, almost anyone can really use AI to generate something. And generative AI has really, really changed the landscape of who can create these AI outputs. However, it's important to note that while anyone can now create and work with AI, that doesn't mean everyone is an AI developer.
Rise of AI “Citizen Developer”
Generative AI has had some very interesting implications for organizations when it comes to AI teams. A few years ago, if you surveyed organizations and asked who was working on AI projects, most people would probably say they weren't working on any. But now, if you randomly poll organizations and ask them the same question, “Who is using AI to do things?”, it might actually be the majority of companies. . Of course, what they're thinking about is quick engineering to write better emails or blog posts or generate slides. Or maybe you're using an image generation tool, and you're incorporating AI into your existing tools and sprinkling them with AI to make them more efficient.
Just as spreadsheets democratized casual quantitative analysis, generative AI is also democratizing casual natural language processing and image generation and analysis tasks. The vast majority of Generative AI users are so-called “citizen AI developers,” much in the same way that no-code and low-code users are empowered as “citizen developers.” Some AI patterns with narrow scope and fast iteration cycles are actually possible using Citizen AI Developers. This is especially true for aspects of conversation patterns, patterns and anomalies, and recognition patterns.
And it's true that many basic AI projects can actually be done by casual users of AI citizen developers. Organizations don't have to compete for highly skilled, highly paid talent for these applications. But even in these cases, you need to manage the quality and availability of data input and output, consider issues of reliable AI, and connect outputs to other systems.
And while these generative AI users may think of themselves as doing things with AI, they don't think of themselves as actual deep AI developers. You cannot build ML models or retrain these models. In short, while AI citizen developers have a place in organizations, they are not replacing AI specialists.
Who are the key players in AI?
AI specialists are still needed because most of the seven patterns of AI cannot be addressed by simple generative AI applications. Patterns such as autonomous, goal-driven, and hyper-personalization require sophisticated, specialized teams and roles.
Let's take self-driving cars as an example. You cannot use a large language model to move a vehicle from point A to point B. So these traditional teams for AI project development are still needed. These teams also consist of data scientists, machine learning engineers, data engineers, and operationalization teams. So when you actually bring AI into the real world, you need a team that can actually do it.
The scope of the AI team is also expanding. The team now includes not only a trusted AI team, but also an AI project management role. These groups are also responsible for aspects of privacy, compliance, risk, ethics, governance, and even intellectual property licensing and management.
Today's AI teams are much broader and more complex than their more “research” counterparts in the past. AI has expanded to become more inclusive. This means that AI is a team effort and everyone in your organization is actually part of the AI team.
The growing role of AI project managers
Technology is just one part of the three-legged stool. His other two lines are people & pProcess and are more important. Too many projects fail by throwing bad technology into a bad problem. Unfortunately, there are still too many solutions looking for problems.
AI project managers and facilitators of all kinds are needed to get an AI project off the ground. These will help you stay focused on your project. This means that AI project management remains extremely important. It is equally important to follow best practices for AI methods. Methodologies such as CPMAI, the Cognitive Project Management for AI method, provide PMs with the step-by-step approach necessary for successful AI projects.
As AI becomes a core part of your organization, everyone will improve, and the more successfully you execute and manage your AI projects, the more your organization will benefit.
(Disclosure: I am a co-host of the AI Today podcast)